Do US green funds perform well?

Universidade do Minho Escola de Economia e Gestão Cátia Catarina Eira Martins Do US green funds perform well? maio de 2020 UMinho | 2020 Cátia Catarina Eira Martins Do US green funds perform well? &iWLD&DWDULQD(LUD0DUWLQV 'R86JUHHQIXQGVSHUIRUPZHOO" 'LVVHUWDomRGH0HVWUDGR 0HVWUDGRHP)LQDQoDV 7UDEDOKRHIHWXDGRVREDRULHQWDomRGD 3URIHVVRUD'RXWRUD0DULDGR&pX&RUWH] 8QLYHUVLGDGHGR0LQKR (VFRODGH(FRQRPLDH*HVWmR PDLRGH ii DIREITOS DE AUTOR E CONDIÇÕES DE UTILIZAÇÃO DO TRABALHO POR TERCEIROS Este é um trabalho académico que pode ser utilizado por terceiros desde que respeitadas as regras e boas práticas internacionalmente aceites, no que concerne aos direitos de autor e direitos conexos. Assim, o presente trabalho pode ser utilizado nos termos previstos na licença abaixo indicada. Caso o utilizador necessite de permissão para poder fazer um uso do trabalho em condições não previstas no licenciamento indicado, deverá contactar o autor, através do RepositóriUM da Universidade do Minho. Licença concedida aos utilizadores deste trabalho Atribuição-NãoComercial-SemDerivações CC BY-NC-ND https://creativecommons.org/licenses/by-nc-nd/4.0/ iii Acknowledgments After a long journey, I have finally finished my dissertation and it is with a great pleasure that I would like to acknowledge all the people, who directly or indirectly, contributed to the achievement of this dissertation. Firstly, my deepest thanks to Professor Maria do Céu Cortez, my supervisor, for all the help, knowledge, information and the availability to answer my doubts. I’m grateful for the guidance and persistent help, without her I wouldn´t be capable to finish this study with success. I would also like to thank all the professors from the Master in Finance, who gave me the tools to perform this study. I also want to thank my colleagues for the extra help and for sharing their knowledge with me. Finally, a special thanks to my family and boyfriend for all the support, for always believing in me and for giving me strength to finish my research. I am very much thankful for their understanding and continuing support to conclude this dissertation. My sincere thanks, to all! iv STATEMENT OF INTEGRITY I hereby declare having conducted this academic work with integrity. I confirm that I have not used plagiarism or any form of undue use of information or falsification of results along the process leading to its elaboration. I further declare that I have fully acknowledged the Code of Ethical Conduct of the University of Minho. v Resumo Os fundos verdes dos EUA têm um bom desempenho? Esta dissertação avalia o desempenho de 13 fundos verdes e 26 fundos convencionais correspondentes dos EUA, de fevereiro de 2004 a setembro de 2019. O desempenho dos fundos é avaliado usando modelos multifatoriais não condicionais, bem como modelos multifatoriais condicionais que permitem que os alfas e betas variem com o tempo. Adicionalmente, este estudo avalia as habilidades dos gestores dos fundos. Além disso, avaliamos também o desempenho dos fundos em diferentes estados do mercado, incluindo uma variável dummy nos modelos multifatoriais. O benchmark de mercado é representado por dois índices: um índice geral de mercado (S&P500) e um índice socialmente responsável (MSCI KLD 400). Os resultados deste estudo estão de acordo com a maioria dos estudos sobre o desempenho dos fundos verdes e sugerem que os investidores verdes podem esperar não serem nem penalizados nem beneficiados por investir em fundos verdes. Os fundos convencionais também apresentam um desempenho neutro em relação ao mercado. Além disso, os resultados indicam que os fundos verdes não apresentam um desempenho diferente dos fundos convencionais. O desempenho semelhante entre estes dois tipos de fundos parece estar relacionado com as boas habilidades de seletividade dos gestores de fundos convencionais, combinadas com as boas habilidades de timing dos gestores de fundos verdes. Usando o modelo condicional de cinco fatores de Fama and French (2015) com a variável dummy, os fundos verdes apresentam um desempenho inferior aos fundos convencionais em períodos de expansão. No entanto, em períodos de recessão, o desempenho dos fundos verdes não varia, enquanto que o desempenho dos fundos convencionais diminui significativamente. Assim sendo, investir em fundos verdes é uma boa forma de reduzir a desvantagem associada a períodos de recessão. Em suma, os investidores convencionais podem investir em fundos verdes para diversificar e proteger as suas carteiras. Palavras-chave: Desempenho de fundos, Finanças sustentáveis, Fundos convencionais, Fundos socialmente responsáveis, Fundos verdes. vi Abstract Do US green funds perform well? This dissertation evaluates the performance of 13 US green funds and 26 US matched conventional funds from February 2004 to September 2019. Fund performance is evaluated using unconditional multi-factor models as well as conditional multi-factor models that allow for timevarying alphas and betas. Besides that, this study assesses fund managers’ abilities. Furthermore, we evaluate fund performance in different market states, by including a dummy variable in the multi-factor models. The market benchmark is proxied by two market indexes: a general market index (S&P500) and a socially responsible index (MSCI KLD 400). The results of this study are in line with the majority of the studies on the performance of green funds and suggest that green investors may expect no superior or inferior returns by investing in green funds. Conventional funds also present a neutral performance compared to the market. Besides that, the results indicate that green funds do not perform different from conventional funds. The similar performance between these two types of funds seems to be linked with the good selectivity abilities of conventional fund managers combined with the good timing abilities of green fund managers. Using the conditional Fama and French (2015) five-factor model with a dummy variable, green funds underperform conventional funds in expansion periods. However, in recession periods the performance of green funds remains while the performance of conventional funds decreases. So, investing in green funds it is a good way to reduce the downside associated with recessions periods. Overall, conventional investors can invest in green funds in order to diversify and protect their portfolios. Keywords: Conventional funds, Fund performance, Green funds, Socially responsible funds, Sustainable finance. vii Index 1. INTRODUCTION ........................................................................................................ 1 2 LITERATURE REVIEW ........................................................................................................ 3 2.1. The nature of the relationship between corporate social and financial performance ......... 3 2.2. The effects of environmental performance on financial performance ............................... 4 2.3. The performance of green funds .................................................................................... 7 2.4. Selectivity and timing abilities....................................................................................... 10 3. METHODOLOGY.............................................................................................................. 12 3.1. Unconditional Models .................................................................................................. 12 3.2. Conditional models ...................................................................................................... 13 3.3. Managerial Abilities Models .......................................................................................... 15 3.4. Fund performance in different market states ................................................................ 16 4. DATA .............................................................................................................................. 18 5. EMPIRICAL RESULTS ...................................................................................................... 24 5.1 Fund performance using unconditional models .............................................................. 24 5.1.1 Unconditional Carhart (1997) four-factor model .......................................................... 24 5.1.2 Unconditional Fama and French (2015) five-factor model ........................................... 27 5.2 Fund performance using conditional models ................................................................. 30 5.2.1 Conditional Carhart (1997) four-factor model ............................................................. 30 5.2.2 Conditional Fama and French (2015) five-factor model ............................................... 33 5.3. Selectivity and timing abilities....................................................................................... 36 5.3.1 The unconditional four-factor version of the Treynor and Mazuy (1966) model ............ 36 5.3.2 The unconditional five-factor version of the Treynor and Mazuy (1966) model ............. 38 5.3.3 The conditional four-factor version of the Treynor and Mazuy (1966) model ................ 40 5.3.4 The conditional five-factor version of the Treynor and Mazuy (1966) model ................. 42 5 only firms with specific characteristics reduce their pollution levels in a profitable way. Stanwick and Stanwick (1998) show that a firm’s corporate social performance is, in fact, affected by the size of the firm, the level of profitability, and the quantity of pollution released by the firm. In the same line, Wahba (2008) demonstrates that the market rewards firms that protect the environment, arguing in favor of a positive impact of corporate environmental responsibility on its market value. In contrast to Friedman (1970), the author argues that the implementation of an environmental management system can improve firm competitive advantages by optimizing resources usage. In this sense, corporate environmental responsibility will not hurt corporate financial performance. Regarding the Japanese market, Nakao et al. (2007) also find that environmental performance has a positive impact on a firm’s financial performance and vice-versa. Even if the initial investment is not based on socially accepted practices, firms can invest the surplus in environmentally friendly practices, technologies and initiatives. Pollution reduction sometimes is considered as a cost burden on the firm and this cost can reduce its competitiveness. Alternatively, it can be viewed as way to increase efficiency, saving money and giving firms a cost advantage. To resolve this paradox, Hart and Ahuja (1996) examine the relationship between emissions reductions and financial performance. The results suggest that it does pay to be green. The operating performance is better only in the following year after the initiation of the efforts to prevent pollution and reduce emissions, whereas it takes about 2 years to affect financial performance. Furthermore, firms with the highest levels of emission are the ones who gain the most. The relationship between corporate environmental performance and corporate financial performance is mostly studied for developed countries. Because of that, Manrique and MartíBallester (2017) analyze this relationship considering the economic development of the market, during a global financial crisis. The findings show that in times of economic crisis, corporate environmental performance has a positive impact on corporate financial performance. However, the effect is weaker for firms in developed countries, where the improvement only occurs in shortterm corporate financial performance, than for companies in emerging and developing countries. Jo et al. (2015) investigate how environmental costs affect the performance of firms in the financial sector from 29 countries. The results show that lowering environmental costs will lead to 6 an increase in financial performance in the long term. Reducing environmental costs brings advantages to the firms, for example, improving production efficiency and competitiveness, a better company reputation and reducing the cost of capital. They also find that the effect of reducing environmental costs differs for different levels of development of the markets and across regions. For well-developed financial markets this has a rapid effect; however for firms in less-developed financial markets the effect is observed in the long term. Konar and Cohen (2001) also analyze this relationship by relating the market value of firms in the S&P500 to objective measures of their environmental performance, instead of subjective environmental performance criteria. The results point out that bad environmental performance is negatively correlated with the intangible asset value of firms. If firms reduce the emissions of toxic chemicals by 10%, the market value of those firms increases $34 million. Yet, the impact of this effect differs across industries. Although Entreat et al. (2014) point out the lack of consensus, integrating the results of 149 studies by meta-analytic analysis they show that for the majority of the studies this relationship is positive. They also argue that this relationship is stronger when the strategic approach underlying corporate environmental performance is proactive. Using meta-analytical techniques and focusing on corporate carbon performance, Busch and Lewandowski (2018) examine “When does it pay to be green?”. The results show that there is a positive relationship between carbon performance and financial performance, implying that companies have an incentive to engage in carbon mitigation measures. At a portfolio level, Derwall et al. (2005) analyze if investing in portfolios of companies with high environmental standards leads to inferior or superior performance. Based on eco-efficiency scores, the results show that from 1995 to 2003 high-ranked portfolios offer higher returns than their low-ranked counterparts. Another set of studies supports the view that firms with high environmental performance contribute negatively to financial performance. For the US market, Cordeiro and Sarkis (1997) demonstrate a negative relationship between environmental proactivism and financial performance, using a different measure of financial performance: security analyst earnings forecasts. They argue that security analysts anticipate lower earnings-per-share in the short-term, 7 around 1 to 5 years, for companies that are more environmentally proactive. Therefore, there are short-term disadvantages of environmental proactivism. Lioui and Sharma (2012) also find a negative relationship, arguing that environmental strengths and concerns are negatively associated with corporate financial performance, measured by ROA and Tobin’s Q. They argue that this relationship is driven by the fact that investors see environmental initiatives as a cost or disadvantage. At a portfolio level, Boulatoff and Boyer (2009) find evidence that green firms underperform comparable Nasdaq firms. The authors also show that green firms have higher volatility. Haan et al. (2012) also observe a negative relationship between corporate environmental performance and stock returns, motivated by the common risks associated with corporate environmental performance. Finally, some studies do not find any relationship between environmental performance and financial performance. Puopolo et al. (2015) show that there is no linear relationship between being green and financial returns, pointing out that the implementation of environmentally friendly standards is new in the marketplace. It is also important to mention that there is even evidence of curvilinear relationship between environmental performance and firm performance, as in Ramanathan (2018). As firms improve their environmental performance, they achieve higher levels of financial performance, but after a certain level of environmental performance, financial performance deteriorates. Pekovic et al. (2018) also state a non-linear relationship, finding an inverted U-shaped relationship, suggesting that there is an optimal level of environmental investment. 2.3. The performance of green funds There are many studies that compare the performance of socially responsible funds with conventional ones or the market. Overall, the majority of the studies find that SRI funds have similar performance to conventional funds and to the market. There are also some studies on the performance of socially responsible funds that focus specifically on funds that screen for environmental criteria – the so-called green funds. As far we know, the first study that evaluates the performance of green funds in the US and German markets is that of White (1995). The results 8 show that US green funds underperform the market, but green funds perform similarly to the market in Germany. Climent and Soriano (2011) examine the financial performance of US green funds compared to conventional funds from 1987 to 2009. They find that US green funds underperform conventional funds, because green funds are subject to higher risks, since they limit the number of investments in which they can invest. However, when focusing in the period from 2001 to 2009, green funds had a similar performance. The authors argue that the initial poor performance of green funds may be explained by their more restricted investment set. Other possible explanation for these results may be that green funds increased in terms of value faster than conventional funds, due to a higher demand. Chang et.al (2012) compare the financial performance of green and conventional funds in the US. The results of this study indicate that green funds underperform conventional ones. Green mutual funds exhibit higher expenses ratios, lower returns and lower risk-adjusted returns. In terms of risk, green mutual funds’ risk seems to be similar to conventional funds, so green investment restrictions in terms of diversification does not engender more risk. Also for the US market, but from 1998 to 2007, Mallett and Michelson (2010) find that US green fund returns are similar to SRI fund and index returns. They argue that the lack of performance differences between green and SRI funds is due to the small period that green funds have been operating. As time passes, more information and data become available and the difference in green funds and SRI funds may grow wider. Adamo et al. (2014) collected a data set of 257 green funds all over the world and evaluated their performance. The authors concluded that green funds have a growing importance and have a positive performance even in recession periods. Controlling for crisis and non-crisis periods, Muñoz et al. (2014) analyze the financial performance of US and European SRI funds. For US SRI funds, the results show that in crisis periods SRI funds have a statistically insignificant performance, whereas in non-crisis market periods US SRI funds underperform the market. The authors argue that green funds perform similarly to other forms of SRI funds. For European SRI funds, the results show that regardless of market conditions, SRI funds show a statistically insignificant performance. 9 Silva and Cortez (2016) also consider the performance of US and European green funds in different market states. Their results show that green funds tend to underperform the benchmark mainly in non-crisis periods and when short-term interest rates are inferior than normal. Besides that, they find that funds certified with a SRI label do not out-or-under perform green funds without the label. However, the number of funds presenting negative performance is higher for non-certified funds. Lesser and Walkshäusl (2016) analyze the financial performance and screening activity of socially responsible, green, and faith-based equity funds for periods of crisis and non-crisis. In crisis periods, the results show that these three types of funds perform similar to their conventional peers and the market. However, during non-crisis periods, green and socially responsible funds tend to underperform. The authors argue that the performance differences are due to the funds’ screening activities since there are performance drivers and reducers for each strategy. For example, social screens lead to the underperformance of socially responsible funds, while energy screens drive the performance of green funds. Ibikunle and Steffen (2017) perform a comparative analysis of European green, black and conventional mutual funds from 1991 to 2014. The results show that green mutual funds underperform relative to conventional funds. However, there is no significant risk-adjusted performance differences between green and black mutual funds. Peculiarly, the green fund’s riskadjusted return improves until the point where there is no difference in the performance of the green and the conventional funds. From 2012 to 2014, green funds begin to outperform their black peers, since fossil energy and natural resources are being replaced by renewable energy. Within the green arena, several studies have focused specifically on funds that invest in renewable energy companies. The number of alternative energy funds have been growing as a result of many countries trying to encourage stakeholders to consider renewable energy sources. Reboredo et al. (2017) evaluate the performance of alternative energy funds in several countries and show that these funds underperform SRI and conventional mutual funds in terms of returns and downside risk protection. Therefore, investors are paying a premium for being green, especially using renewable energies. Marti‐Ballester (2019a) stresses that renewable energy mutual funds channel private resources into climate finance, if managers adopt renewable energy principles in investors’ portfolios. As such, renewable energy mutual funds play an important role as financial instruments. 10 This author analyzes the performance of these funds in Europe over 2007-2018 and compares their financial performance with black energy and conventional funds. She finds that renewable energy funds outperform the energy benchmark but underperform the fossil fuel energy and conventional market benchmarks. Thus, investing in renewable energy funds has a cost for investors when compared with conventional funds. In another study, Marti‐Ballester (2019b) analyzes the financial performance of energy and renewable energy mutual funds in Europe using conditional and unconditional models. Using unconditional models, the results show that renewable energy mutual funds outperform the specific benchmark but underperform the conventional benchmark. Using conditional models, renewable energy funds perform similarly to the market, but underperform their conventional peers using a specialized market benchmark. This author also concludes that fund characteristics such as SRI certification does not affect the financial performance of renewable energy funds. However the expense ratio has a negative effect on financial performance. 2.4. Selectivity and timing abilities It is also important to consider that fund performance can be a result not only fund managers’ selectivity abilities but also their market timing abilities (Muñoz et al. 2014). Stockpicking and market-timing abilities have been widely discussed in the mutual fund literature, especially for conventional funds. Depending on the market or the period analyzed, the empirical evidence shows mixed results. Muñoz et al. (2014) find that European and US global green funds do not exhibit good timing abilities. Furthermore, for the European market Leite and Cortez (2014) do not find differences in terms of timing abilities between SRI funds and their matched conventional funds. Ang et al. (2014) compare SRI funds in Europe and North America, finding market-timing abilities in both regions. In contrast, Ferruz et al. (2010) find negative timing abilities for both SRI and conventional funds in the UK market. For the Swedish market, Leite et al. (2018) show that SRI and conventional fund managers do not present selectivity and timing abilities. Leite and Cortez (2014) argue that SRI funds present different selectivity and timing abilities relative to conventional funds for several reasons. On the one hand, if the screening 11 process generates informational advantages, it can help fund managers to recognize undervalued securities. On the other hand, the limited investment universe restricts fund managers compared to conventional funds, since SRI should follow social investment criteria. The restrictions in terms of selectivity for SRI fund managers, might motive them to be more focused on market timing opportunities. Yet, SRI fund assume a more long-term perspective compared to their conventional peers, being more loyal to the companies in which they invest, which may limit their possibility to explore market timing opportunities. 12 3. METHODOLOGY This chapter presents the methodology used to evaluate fund performance. Starts by presenting the four-factor model of Carhart (1997) and the five-factor model of Fama and French (2015). Then, we present these models in their conditional specification. Then the models to evaluate the timing and selectivity abilities. Finally, we present the models that include a dummy variable to distinguish performance in recessions and expansions periods. 3.1. Unconditional Models Using a market benchmark as the unique risk factor, Jensen (1968) measures performance as the difference between the actual portfolio’s return and the expected risk-adjusted return based on the CAPM. Despite the popularity of Jensen’s (1968) alpha, it has been widely argued that this model is not sufficiently good at explaining the cross-section of expected stock returns (Fama and French, 1993). One of the reasons is that the single-factor model tends to overestimate fund performance (Elton et al., 1993). In fact, multi-factor models have been recognized as much more useful to characterize portfolio returns than a single-factor model (Derwall et al., 2005; Climent and Soriano, 2011). The Carhart (1997) four-factor model includes the original factors of the Fama and French (1993) three-factor model (market, size and book-to-market) with the momentum factor. This model is expressed by the following equation: 𝑟, = 𝛼+ 𝑏𝑟, + 𝑏(𝑆𝑀𝐵)+ 𝑏(𝐻𝑀𝐿)+ 𝑏(𝑀𝑂𝑀)+ 𝜀, (1) where 𝑟, is the excess return of fund p in period t, 𝑟, represents the market’s excess return in period t, 𝑆𝑀𝐵 (small minus big) is the difference in returns between a portfolio of small stocks and a portfolio of large stocks; 𝐻𝑀𝐿 (high minus low) is the difference in returns between a portfolio of high book-to-market stocks and a portfolio of low book-to-market stocks; 𝑀𝑂𝑀 is the difference in the returns of a portfolio of past winners and a portfolio of past losers and 𝑏,𝑏,𝑏 and 𝑏 are the factor coefficients (betas on each of the factors). 13 A more recent model to evaluate the performance is the five-factor model of Fama and French (2015). Using this model, it has become possible to better understand the investment strategies of funds’ managers, since the Fama and French (1993) model does not explain the variation of returns related to the investment and the profitability. We note that there are few studies evaluating fund performance with this model since it is relatively recent. Besides the market, size and book-to-market factors, this model adds two new factors: profitability (𝑅𝑀𝑊) and investment (𝐶𝑀𝐴), and is expressed by the following equation: 𝑟, = 𝛼+ 𝑏𝑟, + 𝑏(𝑆𝑀𝐵)+ 𝑏(𝐻𝑀𝐿)+ 𝑏(𝑅𝑀𝑊)+ 𝑏 (𝐶𝑀𝐴)+ 𝜀, (2) where 𝑅𝑀𝑊 is the difference between the returns on diversified portfolios of stocks with robust and weak profitability and 𝐶𝑀𝐴 is the difference between the returns on diversified portfolios of the stocks of low and high investment firms (conservative and aggressive). It has been widely debated in literature that unconditional models such as those presented so far can produce biased estimates of performance, as these models assume constant expected returns and risk. Considering this limitation, we also apply conditional models to evaluate performance. These models are more robust, as they assume that expected returns and risk vary over time, considering market conditions. 3.2. Conditional models This conditional approach to evaluate the performance allows betas to vary over time as linear functions of a vector of predetermined information variables. These variables represent the public information that is available at time t-1 in order to predict returns at time t. The conditional single-factor model of Ferson and Schadt (1996) is represented by the following equation: 𝑟, = 𝛼+ 𝛽(𝑟. )+ 𝛽  󰆒(𝑧𝑟. )+ 𝜀, (3) where 𝑧 represents a vector of the deviations of 𝑍 from the unconditional values, 𝛽 is an average beta that represents the unconditional mean of the conditional betas, and 𝛽  󰆒 is a vector that measures the response of the conditional betas to the information variables. 14 The conditional single-factor model proposed by Ferson and Schadt (1996) can be viewed as partial conditional model, considering that it only allows betas to vary over time, while assuming that alphas are constant. Christopherson et al. (1998) extend the model of Ferson and Schadt (1996) by also allowing alphas also to be time-varying, as follows: 𝛼(𝑍)= 𝛼 + 𝑧𝐴  󰆒 (4) where 𝛼 represents the average alpha and 𝐴  󰆒 measures the sensitivity of the conditional alpha to the information variables. Rearranging the equations by combining equations (3) and (4), we have the full conditional of Christopherson et al. (1998), represented by the following equation: 𝑟, = 𝛼 + 𝑧𝐴  󰆒+ 𝛽(𝑟. )+ 𝛽  󰆒(𝑧𝑟. )+ 𝜀, (5) The conditional Carhart (1997) four-factor model is obtained combining the conditional single-factor model with the Carhart (1997) risk factors. Combining equation (5) with the four risk factors gives the conditional multi-factor model with time-varying alphas and betas, represented as follows: 𝑟, = 𝛼 + 𝑧𝐴  󰆒+ 𝛽𝑟. + 𝛽  󰆒(𝑧𝑟. )+ 𝛽,𝑟. + 𝛽 , 󰆒(𝑧𝑟. ) + 𝛽,𝑟. +𝛽 , 󰆒(𝑧𝑟. ) + 𝛽,𝑟.+ 𝛽 , 󰆒(𝑧𝑟. )+ 𝜀, (6) In turn, the conditional Fama and French (2015) five-factor model with time-varying alphas and betas is represented by the following equation: 𝑟, = 𝛼 + 𝑧𝐴  󰆒+ 𝛽𝑟. + 𝛽  󰆒(𝑧𝑟. )+ 𝛽,𝑟. + 𝛽 , 󰆒(𝑧𝑟. ) + 𝛽,𝑟. +𝛽 , 󰆒(𝑧𝑟. ) + 𝛽,𝑟.+ 𝛽 , 󰆒(𝑧𝑟. ) + 𝛽,𝑟. + 𝛽 , 󰆒(𝑧𝑟. )+ 𝜀, (7) 21 Table 1List of US green and conventional funds Fund Name Inception date Lipper Global Classification Total net assets 1 Green ARIEL FOCUS FD.INVR.CL. 01/02/2006 Equity US 40,8 Conventional FID.ADVI.ASST.MANAGER 85% CL.C 02/10/2006 Equity US 40,9 HARTFORD GW.OPPS.FD.CL. R3 21/12/2006 Equity US 44,5 2 Green ASPIRATION REDWOOD FUND 16/11/2015 Equity US 84,5 Conventional SEI INST MGD TAX-MANAGED MANAGED VOLATILITY Y 30/04/2015 Equity US 87,7 AQR TM LARGE CAP MULTISTYLE FUND I 11/02/2015 Equity US 92,8 3 Green BROWN ADVISORY WINSLOW SUSTBY.FD.INSTL.SHS. 29/06/2012 Equity US 861,3 Conventional TOUCHSTONE FOCD.FD.CL.Y 16/04/2012 Equity US 829,1001 WELLS FARGO PREMIER LARGE CO GR FD R6 03/12/2012 Equity US 789,5 4 Green PARNASSUS ENDEAVOR FUND INVESTOR 29/04/2005 Equity US 2709,7 Conventional AMERICAN CENTURY MID CAP VALUE FUND I 31/01/2005 Equity US 2375,5 MFS VAL.FD.CL.R4 01/04/2005 Equity US 2975,6 5 Green PARNASSUS FUND INVESTOR 27/08/1987 Equity US 776,1001 Conventional ANCHOR SA WELLINGTON CAPITAL APPRECTN PORT 1 23/03/1987 Equity US 771 AMG MANAGERS BRANDYWINE FUND I 23/05/1986 Equity US 786,2 6 Green PARNASSUS MID CAP FUND INVESTOR 29/04/2005 Equity US 2271,2 Conventional ADVANCED SRS WELLINGTON MANAGEMENT HEDGED EQ PTF 05/12/2005 Equity US 2067,6 DFA US.CORE EQ.1 PRTF. 21/10/2005 Equity US 26071,1 7 Green TIAA-CREF INSTL.SOCIAL CHOICE EQ.FD.RTMT.CL. 12/12/2002 Equity US 657,7 Conventional AMERICAN FUNDS AMCAP FUND R2 31/05/2002 Equity US 614,8 BRIGHTHOUSE/WELL CORE EQUITY OPPTY PTFL B 30/07/2002 Equity US 683,2 8 Green GREEN CENTURY EQUITY FUND INDIVIDUAL INVESTOR 22/09/1997 Equity US 243,4 Conventional AB GROWTH FUND ADVISOR 31/03/1997 Equity US 202,8 ALGER CAP.APPREC.FD.CL.C 08/08/1997 Equity US 217,3 9 Green PARNASSUS CORE EQUITY FUND INVESTOR 19/04/1993 Equity US Income 9591,898 Conventional PTNM.EQ.INC.FD.CL.A 14/10/1993 Equity US Income 8184,898 JP MORGAN EQUITY INCOME FUND I 01/09/1989 Equity US Income 9908 10 Green ARIEL APPRECIATION FUND INVESTOR CL. 19/07/1990 Equity US Sm&Mid Cap 997,3 Conventional ICM SML.CO.PRTF.INSTL. CL. 28/02/1991 Equity US Sm&Mid Cap 890 INVESCO OPPENHEIMER MID CAP VALUE FUND A 03/12/1991 Equity US Sm&Mid Cap 824,2 11 Green ARIEL FUND INVESTOR CL. 16/03/1987 Equity US Sm&Mid Cap 1303,8 Conventional NORTHWESTERN MUTUAL MCG STK PFOLIO 30/11/1990 Equity US Sm&Mid Cap 1095,2 INVESCO OPPENHEIMER DISCOVERY FUND A 02/02/1987 Equity US Sm&Mid Cap 1341,1 12 Green WALDEN SMALL CAP FUND 30/10/2008 Equity US Sm&Mid Cap 110,5 Conventional MML MID CAP GROWTH FUND SERVICE 15/08/2008 Equity US Sm&Mid Cap 103,4 NATIONWIDE NVIT MULTMNGR MCG FD II 24/03/2008 Equity US Sm&Mid Cap 138,2 13 Green WALDEN SMID CAP FUND 29/06/2012 Equity US Sm&Mid Cap 57,1 Conventional NATIONWIDE BAILARD COGNITIVE VALUE FUND M 16/09/2013 Equity US Sm&Mid Cap 60 MADISON MID-CAP FD.CL.R6 29/02/2012 Equity US Sm&Mid Cap 54,1 This table presents the sample of green funds identified using the US SIF 2018 and the respective conventional funds. For each fund present the name of the fund, inception date, Lipper Global classification and the total net assets extracted from DataStream. 22 Over the period under analysis both fund portfolios and the market factors present positive mean excess returns, except the book-to-market (HML) and investment factor (CMA). Comparing the green fund portfolio with the green benchmark, the green portfolio presents higher mean monthly excess returns and a higher standard deviation. In addition, the conventional fund portfolio presents higher mean monthly excess returns and higher standard deviation than the conventional benchmark. Comparing both portfolios, the green portfolio presents higher mean excess returns than conventional funds and also a higher standard deviation, meaning that green funds present a higher risk. Comparing the benchmarks, the conventional benchmark has a higher mean monthly excess return and a lower standard deviation than the benchmark of the sector. Regarding the symmetry of the distribution, both portfolios and benchmarks have a negative skewness (negatively skewed), which indicates that the left tail of the distribution is greater than the right tail. Regarding to the characterization of the peak of the distribution, both portfolios and benchmarks have exhibit excess kurtosis (higher than 3), which classifies it as leptokurtic. Additionally, the normality test was performed. The results support that the portfolios of green and conventional funds and the benchmarks are not normally distributed, since we do not accept the null hypothesis of normality at the level of 5 %. For the risk factors, we only reject the null hypothesis for the HML and MOM factors. As argued by Adcock et al. (2012), the rejection of normality of returns supports the application of conditional models. Table 2Descriptive statistics of US green and conventional funds, market benchmarks and risk factors No. of obs. Mean excess returns (%) Standard deviation (%) Kurtosis Skewness Min Max Adj. 𝝌 𝟐 P value US green portfolio 188 0.672 4.44 5.866 -0.474 -0.196 0.175 19.73 0.0001 US conventional portfolio 188 0.663 4.20 5.192 -0.828 -0.187 0.117 24.90 0.0000 S&P500 188 0.662 3.93 5.090 -0.764 -0.169 0.109 15.72 0.0004 MSCI KLD 400 188 0.644 3.94 4.475 -0.594 -0.155 0.106 22.81 0.0000 SMB 188 0.0422 2.38 2.837 0.303 -0.0478 0.0681 3.08 0.2148 HML 188 -0.0737 2.55 5.364 0.0765 -0.112 0.0829 12.55 0.0019 RMW 188 0.307 1.55 3.468 0.240 -0.0399 0.0508 3.86 0.1449 CMA 188 -0.0434 1.41 2.873 0.329 -0.0333 0.0370 3.54 0.1699 MOM 188 0.109 4.40 22.29 -2.659 -0.344 0.125 . 0.0000 This table reports summary statistics for equally weighted portfolios of US green and conventional funds, market benchmarks and the additional risk factors. Mean excess returns, standard deviation, kurtosis, skewness, minimum and maximum for the period of February 2004 to September 2019. The Adj. 𝝌 𝟐 is a statistic that is around a 𝝌 𝟐 distribution with 2 degrees of freedom under the null of normality. The result “.” should be interpreted as an absurdly large number so the data are most surely not normal. 23 Table 3 reports the monthly mean excess returns for both portfolios (green and conventional) by years, from 2004 to 2019. From this table, it seems that green and conventional funds present similar fluctuations over time. In 2008, the year of the global financial crisis, the mean monthly excess returns dropped drastically for both green and conventional funds. This drop was higher for conventional funds. The mean monthly excess returns start to increase again in 2009. Analyzing the expansion periods, overall, conventional funds present higher values for the mean excess returns. However, in recession periods, green funds present higher values. Yet, there are no statistically significant differences between the mean monthly excess returns of green and conventional funds. These results suggest the importance of controlling the funds’ performance by market states, since these fluctuations may be in accordance to the business cycles identified by the NBER. Table 3Mean excess returns by years for green fund and conventional funds Year Green (1) (%) Conventional (2) (%) Difference (1)-(2) (%) p-value 2004 0.694 0.814 -0.1202 0.9205 2005 -0.076 0.458 -0.534 0.6218 2006 0.703 0.705 -0.002 0.9985 2007 -0.0898 0.601 -0.691 0.5602 2008 -3.347 -3.912 0.565 0.8472 2009 3.262 2.46 0.803 0.7755 2010 1.464 1.588 -0.124 0.9601 2011 -0.0652 -0.1472 0.0820 0.9706 2012 1.359 1.256 -0.103 0.9410 2013 2.571 2.464 0.107 0.9202 2014 0.843 0.801 0.042 0.9719 2015 -0.0898 0.160 -0.25 0.8700 2016 1.245 0.793 0.453 0.7554 2017 1.315 1.531 -0.2156 0.6721 2018 -0.728 -0.571 -0.157 0.9360 2019 2.024 1.945 0.0773 0.9730 This table reports the mean excess returns for the equally weighted portfolios of green and conventional, and for the difference between these two, by years. The p value is calculated for the difference of the mean between green and conventional funds. 24 5. EMPIRICAL RESULTS This chapter presents the results on the performance of the US green and conventional funds. The analysis starts with the results of the unconditional multi-factor models (Carhart, 1997 and Fama and French, 2015) and then those of the conditional approach, as in Christopherson et al. (1998), applied to each model. Then we compare timing and selectivity abilities of both types of funds, using the multifactor version of the Treynor and Mazuy (1966) model. To finish we present the results of the Carhart (1997) four-factor model and Fama and French (2015) with a dummy variable, in order to analyze the performance of green and conventional funds in different market states. 5.1 Fund performance using unconditional models Considering that multi-factor models are more useful to explain the cross-section of expected stock returns, we apply the Carhart (1997) four-factor model and the Fama and French (2015) five-factor model to evaluate fund performance. 5.1.1 Unconditional Carhart (1997) four-factor model Table 4 presents the results of fund performance at the aggregate level for the Carhart (1997) four-factor model from 2004 to 2019 and summarizes the results on individual fund performance. Appendixes 1 and 2 detail the results for individual funds. Panels A and B show the results considering a conventional index (S&P500) and a green index (MSCI KLD 400), respectively, as the market benchmark. As in Climent and Soriano (2011), this study also evaluates the “difference” portfolio, constructed by subtracting the returns of the conventional portfolio from the green portfolio. This portfolio is constructed in order to evaluate the differences in risk and return between the different investment approaches. The explanatory power of the models is above 95% for the green and conventional portfolio regressions, which means that more than 95% of the variability of the excess returns is explained by the model. The explanatory power is slightly higher when the benchmark is the S&P500, which 25 means that the conventional index is more capable to explain portfolio performance than the SRI index. Analysing panel A, none the portfolio alphas are statistically significant. Individually, there are 5 green funds with positive alpha coefficients, but only one is statistically significant. Regarding conventional funds, there are 10 conventional funds with positive alpha coefficients, but only one is statistically significant. Overall, the majority of green and conventional funds present neutral performance, so we can conclude that neither green or conventional funds perform significantly differently from the market. These results are consistent with previous studies (e.g., Climent and Soriano, 2011; Muñoz et al., 2014) on the performance of green funds. Climent and Soriano (2011) show that green funds underperform the market benchmark from 1987 to 2001, which is not in accordance with panel A. However, focusing in the period from 2001 to 2009, Climent and Soriano (2011) show that green and conventional funds did not perform differently from the market Table 4 - Unconditional Carhart four-factor model performance Panel A: Benchmark S&P500 Portfolios 𝜶 𝒑 𝜷 𝒑 𝜷 𝑺𝑴𝑩 𝜷 𝑯𝑴𝑳 𝜷 𝑴𝑶𝑴 𝑹 𝟐 adj. Green (1) 0.0002 0.9910*** 0.3075*** -0.0022 -0.1083*** 97.44% N+ N5[1] 7[0] 13[13] 0[0] 13[12] 0[0] 4[3] 9[3] 4[1] 9[7] Conventional (2) -0.0003 1.0030*** 0.3269*** -0.1417*** 0.0264* 97.99% N+ N10[1] 16[4] 26[26] 0[0] 23[22] 3[0] 10[6] 16[12] 13[6] 13[5] Difference (1)-(2) 0.0005 -0.0120 -0.0193 0.1364*** -0.1347*** 45.77% Panel B: Benchmark MSCI KLD 400 Portfolios 𝜶 𝒑 𝜷 𝒑 𝜷 𝑺𝑴𝑩 𝜷 𝑯𝑴𝑳 𝜷 𝑴𝑶𝑴 𝑹 𝟐 adj. Green (1) 0.0003 1.0026*** 0.2527*** 0.0340 -0.0849*** 97.23% N+ N6 [1] 7[1] 13[13] 0[0] 12[10] 1[1] 7[4] 6[3] 6[1] 7[7] Conventional (2) -0.0000 1.0000*** 0.2791*** -0.1015** 0.0463** 95.68% N+ N10[1] 14[2] 26[26] 0[0] 22[17] 4[1] 10[7] 16[12] 16[7] 10[2] Difference (1)-(2) 0.0004 0.0026 -0.0264 0.1355*** -0.1312*** 45.64% This table presents regression estimates for the equally weighted portfolios of US green and conventional funds, as well as the difference between these two portfolios, obtained from the four-factor model regressions with both S&P500 (Panel A) and KLD400 (Panel B) as benchmarks, from February 2004 -September 2019. It reports estimates of performance (𝜶𝒑), systematic risk (𝜷𝒑) , factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination (𝑅 adj. ). Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). N+ and Nindicate the number of the funds that have positive and negative estimates, respectively. Within brackets the number of funds whose estimates are statistically significant at a 5% significance level are presented. 26 and that the difference between green and conventional funds is statistically insignificant. In panel B, using the MSCI KLD 400 as the market benchmark, the alpha coefficients are also statistically insignificant for both green and conventional portfolios, meaning a neutral performance. In sum, comparing the performance of green and conventional funds, whatever benchmark is used, we can observe that the alphas of the green performance are higher than those of the conventional portfolios, although the difference is not statistically significant. In relation to market risk, the results show that betas are statistically significant at the 1% level for both green and conventional portfolios. Green funds present higher values of beta when the benchmark is the green benchmark (MSCI KLD 400), whereas the conventional funds exhibit higher beta when the benchmark is the conventional benchmark (S&P500). This means that green funds are more exposed to the green benchmark and the conventional funds to the conventional benchmark. However, there is no statistically significant difference regarding the market risk between the two portfolios. At the individual level, all coefficients associated with the market risk are positive and statistically significant. Regarding the risk factors, the results demonstrate that green and conventional funds are more exposed to the size (SMB) factor, since this risk factor is statistically significant at the 1% level. At the individual level, the majority of the funds also present positive and statistically significant SMB coefficients. The betas associated to the book-to-market (HML) factor are only statistically significant for conventional funds, with a negative sign, suggesting that conventional funds are more exposed to growth stocks than to value stocks. Observing the coefficients of the difference portfolio, we conclude that green funds are more exposed to value stocks than conventional funds. The MOM risk factor is statistically significant for all portfolios. However, the green portfolio presents negative coefficients and the conventional portfolio positive coefficients. This means that green funds are more exposed to companies with poor performance in the recent past, while conventional funds are more exposed to companies with a good past performer. 27 5.1.2 Unconditional Fama and French (2015) five-factor model The Fama and French (2015) model adds the profitability (RMW) and the investment (CMA) factors, excluding the momentum (MOM) factor, to the previous model. The results of this model are presented in table 5. Detailed estimates on the individual funds are reported in Appendixes 3 and 4. Compared with the previous model, we observe that in spite of adding the two risk factors, fund performance estimates remains neutral. In fact, the alpha coefficients remain statistically insignificant, being consistent with the previous results. The difference between the performance of the green portfolio and the conventional portfolio is not statistically significant. At the individual level, in panel A one green fund and two conventional funds outperform the market. Additionally, two conventional funds underperform. In panel B, two green funds and Table 5Unconditional Fama and French (2015) five-factor model performance Panel A: Benchmark S&P500 Portfolios 𝜶 𝒑 𝜷 𝒑 𝜷 𝑺𝑴𝑩 𝜷 𝑯𝑴𝑳 𝜷 𝑹𝑴𝑾 𝜷 𝑪𝑴𝑨 𝑹 𝟐 adj. Green (1) -0.0002 1.0246*** 0.3088*** 0.0637** 0.0237 0.0004 96.53% N+ 5[1] 13[13] 13[12] 8[4] 8[2] 7[0] N8[0] 0[0] 0[0] 5[1] 5[0] 6[1] Conventional (2) 0.0003 0.9724*** 0.3032*** -0.1237*** -0.1275*** -0.1154*** 98.20% N+ 13[2] 26[26] 23[20] 11[9] 8[5] 6[3] N13[2] 0[0] 3[0] 15[11] 18[11] 20[11] Difference (1)-(2) -0.0005 0.0522*** 0.0056 0.1874*** 0.1512*** 0.1157* 29.02% Panel B: Benchmark MSCI KLD 400 Portfolios 𝜶 𝒑 𝜷 𝒑 𝜷 𝑺𝑴𝑩 𝜷 𝑯𝑴𝑳 𝜷 𝑹𝑴𝑾 𝜷 𝑪𝑴𝑨 𝑹 𝟐 adj. Green (1) -0.0000 1.0293*** 0.2537*** 0.1048*** 0.0323 -0.0566 96.71% N+ 5 [2] 13 [13] 12 [8] 8 [5] 8[2] 3[6] N8 [1] 0[0] 1[1] 5[2] 5[0] 10[2] Conventional (2) 0.0006 0.9579*** 0.2574*** -0.0781* -0.1338*** -0.1795*** 95.93% N+ 15 [2] 26 [26] 22 [16] 11[9] 7 [5] 4 [1] N10 [1] 0 [0] 4 [1] 15[11] 19[11] 22 [12] Difference (1)-(2) -0.0007 0.0714*** -0.0037 0.1830*** 0.1661*** 0.1229** 31.15% This table presents regression estimates for the equally weighted portfolios of US green and conventional funds, as well as the difference between these two portfolios, obtained from the five-factor model regressions with both S&P500 (Panel A) and KLD400 (Panel B) as benchmarks, from February 2004 - September 2019. It reports estimates of performance (𝜶𝒑), systematic risk (𝜷𝒑), factor loadings associated to size (SMB), profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination (𝑅 adj.) . Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). N+ and Nindicate the number of the funds that have positive and negative estimates, respectively. Within brackets the number of funds whose estimates are statistically significant at a 5% significance level are presented. 28 two conventional funds present a positive and statistically significant alpha, meaning that these funds outperform the market, and only one green and one conventional funds underperform the market. Overall, the majority of the funds present a neutral performance. Comparing the 𝑅 adj. of both models, the five-factor model presents a slightly lower 𝑅 adj. for green funds and a slightly higher for conventional funds, consistent with, the two additional factors being statistically insignificant for green funds. In terms of market exposures, all betas are positive and statistically significant at the 1% level for both green and conventional funds. As in the previous model, individually all the coefficients (of green and conventional funds), are statistically significant. However, with this model we observe a statistically significant difference in terms of market exposure, as green funds are more exposed to the market in comparison to conventional funds. Analyzing the additional risk factors of this model, the size (SMB) factor is still the most relevant one. All betas associated with this risk factor are positive and statistically significant for both green and conventional funds, regardless of the benchmark. This means that both green and conventional funds are more exposed to small caps, in line with the results of four-factor model. At the individual fund level, the majority of the funds show a positive and statistically significant coefficient associated with the size (SMB) factor. As in previous results, there is no statistically significant difference between the two portfolios in terms of the size coefficient. The betas of the book-to-market (HML) factor are positive and statistically significant for green funds, indicating that green funds are more exposed to value stocks. Though, for conventional funds the betas of the HML factor are negative and statistically significant, meaning that conventional funds are more exposed to growth stocks than to value stocks, as in the previous model. Besides that, the book-to-market factor associated with the difference portfolio is positive and statistically significant, meaning that green funds are more exposed to value stocks than conventional funds, also in line with the previous model. The two additional risk factors, the profitability (RMW) and the investment (CMA) factors, do not present significant coefficients for green funds. However, for conventional funds, these two factors are statistically significant and negative. This means that conventional funds are more exposed to companies with weak profitability and high investment firms. Furthermore, looking at the difference portfolio, the betas associated with these two additional factors are positive and 29 statistically significant, meaning that green funds are more exposed to firms with robust profitability and low investments than conventional funds. 30 5.2 Fund performance using conditional models The conditional approach proposed by Ferson and Schadt (1996) only allows betas to be time-varying but alpha remains constant. The full conditional approach of Christopherson et al. (1998) allows for time-varying betas and alphas. As argued by Ferson et al. (2008), using the model that allows for betas to vary over time but forcing alphas to be constant will generated bias results. So, as mentioned in section 3, the full conditional specification of the multi-factor models is applied. In this model two public information variables are used: the short-term rate (ST) and the dividend yield (DY). 5.2.1 Conditional Carhart (1997) four-factor model Table 6 presents the results of the full conditional four-factor model of Carhart (1997). Appendixes 5 and 6 contain the detailed results for individual funds. Analyzing the 𝑅𝑠 adj. of the models, they are not much different from those obtained with the unconditional approach. However, the 𝑅 s adj. are now slightly higher, so we conclude that the explanatory power of the conditional models increases in relation to the unconditional models, demonstrating the importance of including public information variables. The results show that both green and conventional portfolios present a neutral performance compared to the market, as in the previous findings. Analyzing the difference portfolio, there is no statistically significant difference between the performance of green and conventional funds. Thus, US green funds do not perform differently from conventional funds. At the individual level, in panel A, none of the green or conventional funds outperform the market. However, four green funds and three conventional funds perform worse than the market. In panel B, one green fund outperforms the market benchmark, but two green and two conventional funds underperform the market. Overall, the majority of the funds present a neutral performance. In panel A, the alphas associated with the dividend yield and the short-term rate present a neutral influence in explain the performance of green and conventional funds. However, in panel B, the alpha associated with the dividend yield present a negative and statistically significant value at 10% level, meaning that green funds present a lower performance in times of higher dividends. 37 Overall the results suggest that there are no statistically significant differences in terms of stock-picking abilities between green and conventional managers. In terms of differences to time the investments styles, conventional fund managers are less skilled in timing the momentum style than green fund managers. Table 8 – The unconditional four-factor version of the Treynor and Mazuy (1966) model Panel A: S&P500 Portfolios Green (1) NN+ Conventional (2) NN+ Difference (1)-(2) 𝛼  -0.0002 9[1] 4[1] 0.0004 13[0] 13[2] -0.0006 𝛽  ∗  1.0010*** 0[0] 13[13] 1.0000*** 0[0] 26[26] 0.0011 𝛽  ∗   -0.1266 5[3] 8[0] -0.3399** 18[5] 8[0] 0.2133 𝛽  0.2997*** 0[0] 13[11] 0.3317*** 3[0] 23[22] -0.0320 𝛽   -0.8193 8[1] 5[0] -1.0044** 19[4] 7[0] 0.1851 𝛽  0.0071 8[3] 5[3] -0.1431*** 16[12] 10[7] 0.1502*** 𝛽   0.6586** 4[0] 9[3] 0.4546 10[0] 16[2] 0.2040 𝛽  -0.0726*** 9[6] 4[0] 0.0304* 13[3] 13[5] -0.1031*** 𝛽   0.2630*** 3[0] 10[5] 0.0639 10[0] 16[4] 0.1991** 𝑹 𝟐 adj. 97.55% 98.03% 46.76% Panel B: MSCI KLD 400 Portfolios Green (1) NN+ Conventional (2) NN+ Difference (1)-(2) 𝛼  0.0006 5[1] 8[2] 0.0015* 6[0] 20[3] -0.0009 𝛽  ∗  1.0075*** 0[0] 13[13] 0.9911*** 0[0] 26[26] 0.0164 𝛽  ∗   -0.5323 11[2] 2[0] -0.8174** 24[12] 2[0] 0.2851* 𝛽  0.2459*** 1[1] 12[9] 0.2859*** 4[1] 22[17] -0.0400 𝛽   -1.0774* 11[1] 2[0] -1.2716 19[5] 7[0] 0.1942 𝛽  0.0413* 5[3] 8[3] -0.1100** 16[12] 10[7] 0.1513*** 𝛽   0.7031* 3[0] 10[3] 0.4129 8[0] 18[1] 0.2903 𝛽  -0.0509*** 7[6] 6[2] 0.0466* 12[1] 14[5] -0.0975*** 𝛽   0.3098*** 2[0] 11[7] 0.1159 7[3] 19[5] 0.1938** 𝑹 𝟐 adj. 97.39% 95.88% 47.05% This table presents regression estimates for the equally weighted portfolios of US green and conventional funds, as well as the difference between these two portfolios, obtained by from the Treynor and Mazuy (1966) extended to a multifactor setting regressions with both S&P500 (Panel A) and KLD400 (Panel B) as benchmarks, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability (𝜶𝒑), systematic risk (𝜷𝒑), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination (𝑅 adj) . rm2, SMB2, HML2 and MOM2 refers to squared risk factors. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). N+ and Nindicate the number of the funds that have positive and negative estimates, respectively. Within brackets the number of funds whose estimates are statistically significant at a 5% significance level are presented. . 38 5.3.2 The unconditional five-factor version of the Treynor and Mazuy (1966) model Table 9 presents the results for the Treynor and Mazuy (1966) extended to a five-factor setting. Appendixes 11 and 12 detail the results for individual funds, for the full period time, 20042019. The results suggest that conventional fund managers have a successful stock-picking ability, while green fund managers do not present this ability. Additionally, the differences portfolio shows that conventional fund managers have better selectivity abilities than green fund managers. At the individual level, the majority of the green and conventional funds exhibit neutral selectivity abilities. It is worth noting that although at the portfolio level conventional fund managers present this security selection skills, at the individual fund level only 3 funds present a positive and statistically significant alpha. The portfolio results seem to be driven by these three funds. The type of evidence reinforces the relevance of analyzing performance at the individual fund level to complement the analysis at the aggregate level. With respect to investment styles the only successful style-timing ability for green fund managers is the ability to time the book-to-market style. Besides that, there is a statistically significant difference between the two portfolios, as green fund managers are better in timing the book-to-market style than conventional fund managers. At the individual level, most of the conventional and green funds managers do not present selectivity and timing abilities. And although conventional fund managers present better selectivity abilities, green fund managers are better in timing, namely timing the book-to-market style. 39 Table 9 – The unconditional five-factor version of the Treynor and Mazuy (1966) model Panel A: S&P500 Portfolios Green (1) NN+ Conventional (2) NN+ Difference (1)-(2) 𝛼  -0.0008 8[0] 5[1] 0.0014** 8[0] 18[3] -0.0023* 𝛽  ∗  1.0391*** 0[0] 13[13] 0.9654*** 0[0] 26[26] 0.0737*** 𝛽  ∗   0.2894 4[2] 9[0] -0.1809 15[3] 11[0] 0.4703 𝛽  0.2937*** 0[0] 13[10] 0.3108*** 3[0] 23[21] -0.0171 𝛽   0.2677 6[0] 7[0] -0.7008 15[3] 11[0] 0.9685 𝛽  0.0861** 5[1] 8[4] -0.1206*** 15[11] 11[7] 0.2067*** 𝛽   1.6497*** 2[0] 11[6] 0.3773 13[0] 13[3] 1.2724** 𝛽  0.0335 5[0] 8[2] -0.0873*** 18[7] 8[5] 0.1208** 𝛽   -1.4061 9[0] 4[0] -3.7887*** 20[6] 6[0] 2.3826 𝛽  -0.0344 9[1] 4[0] -0.1130*** 19[9] 7[2] 0.0786 𝛽   -4.1600 9[4] 4[0] 0.9402 9[1] 17[1] -5.1001 𝑹 𝟐 adj. 96.72% 98.30% 32.94% Panel B: MSCI KLD 400 Portfolios Green (1) NN+ Conventional (2) NN+ Difference (1)-(2) 𝛼  -0.0005 10[1] 3[1] 0.0020** 7[0] 19[3] -0.0026** 𝛽  ∗  1.0389*** 0[0] 13[13] 0.9494*** 0[0] 26[26] 0.0895*** 𝛽  ∗   -0.0285 7[1] 6[0] -0.6049* 21[4] 5[0] 0.5764 𝛽  0.2503*** 1[1] 12[7] 0.2770*** 3[1] 23[16] -0.0268 𝛽   -0.0894 7[0] 6[0] -0.9336 18[3] 8[0] 0.8442 𝛽  0.1134*** 5[2] 8[4] -0.0919* 15[10] 11[7] 0.2053*** 𝛽   1.3692** 3[0] 10[3] 0.0890 15[1] 11[3] 1.2801** 𝛽  0.0449 4[0] 9[1] -0.0848* 18[9] 8[5] 0.1297*** 𝛽   -0.7985 7[1] 6[0] -3.3919* 20[3] 6[0] 2.5933 𝛽  -0.0804* 10[2] 3[0] -0.1589** 21[10] 5[2] 0.0785 𝛽   -1.1092 6[1] 7[1] 3.9388 2[0] 24[4] -5.0481 𝑹 𝟐 adj. 96.72% 96.15% 35.95% This table presents regression estimates for the equally weighted portfolios of US green and conventional funds, as well as the difference between these two portfolios, obtained by from the Treynor and Mazuy (1966) extended to a unconditional multifactor setting regressions with both S&P500 (Panel A) and KLD400 (Panel B) as benchmarks, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability (𝜶𝒑), systematic risk (𝜷𝒑), factor loadings associated to size (SMB), book-to-market (HML), profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination (𝑅 adj) . rm2, SMB2 HML2, RMW2 and CMA2 refers to squared risk factors. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). N+ and Nindicate the number of the funds that have positive and negative estimates, respectively. Within brackets the number of funds whose estimates are statistically significant at a 5% significance level are presented. . 40 5.3.3 The conditional four-factor version of the Treynor and Mazuy (1966) model Table 10 presents the results for the Treynor and Mazuy (1966) extended to a conditional four-factor setting. Appendixes 13 and 14 detail the results for individual funds, for the full period time (2004-2019). Regarding fund managers stock-picking ability, green fund managers do not present this ability, while conventional fund managers present a successful stock-picking ability, in the case of the MSCI KLD 400. Besides, there is a little evidence (only at the 10% level), that conventional fund managers are better in terms of selectivity than green fund managers (panel B). Regarding the coefficient of the market squared risk factor, it is statistically insignificant for the green portfolio, while for the conventional portfolio is negative and statistically significant in Panel B. This means that conventional fund managers time the market incorrectly. It is worth mentioning that at the individual level, the number of conventional funds with a significant negative market timing coefficient decreases compared to the unconditional model. This is consistent with Ferson and Schadt (1996) who claim that negative timing observed when using unconditional models tends to decrease when using models that allow for time-varying risk. In relation to investment styles, the only successful style-timing ability is the ability of green fund managers to time the momentum style, in the case of the MSCI KLD 400. The conventional portfolio presents a negative and statistically significant coefficient associated with the size and book-to-market squared risk factors, meaning that conventional fund managers time these styles but in the wrong direction. Additionally, the difference portfolio shows that green fund managers are better than conventional fund managers in timing the book-to-market style, as in the results presented in tables 8 and 9. 41 Table 10 – The conditional four-factor version of the Treynor and Mazuy (1966) model Panel A: S&P500 Portfolios Green (1) NN+ Conventional (2) NN+ Difference (1)-(2) 𝛼  -0.0004 8[1] 5[1] 0.0008 8[1] 18[2] -0.0012 𝛽  ∗  1.0030*** 0[0] 13[13] 0.9820*** 0[0] 26[26] 0.0210 𝛽  ∗   -0.0384 7[3] 6[0] -0.1411 15[4] 11[1] 0.1027 𝛽  0.3022*** 0[0] 13[10] 0.3419*** 3[0] 23[22] -0.0397 𝛽   -0.5294 8[1] 5[0] -1.1050** 19[3] 7[0] 0.5756 𝛽  -0.0043 7[3] 6[3] -0.1053*** 14[12] 12[8] 0.1009*** 𝛽   0.4594 4[0] 9[1] -0.6779** 18[5] 8[2] 1.1373* 𝛽  -0.0698*** 9[5] 4[1] 0.0384** 11[2] 15[5] -0.1082*** 𝛽   0.1007 7[1] 6[2] 0.0204 14[3] 12[4] 0.0803 𝑹 𝟐 adj. 97.83% 98.32% 56.08% Panel B: MSCI KLD 400 Portfolios Green (1) NN+ Conventional (2) NN+ Difference (1)-(2) 𝛼  0.0007 4[1] 9[1] 0.0020** 4[0] 22[5] -0.0013* 𝛽  ∗  1.0017*** 0[0] 13[13] 0.9716*** 0[0] 26[26] 0.0301 𝛽  ∗   -0.4207 10[3] 3[0] -0.6236*** 22[7] 4[0] 0.2029 𝛽  0.2560*** 1[1] 12[9] 0.2958*** 4[1] 22[20] -0.0398 𝛽   -0.8058 9[1] 4[0] -1.2475* 19[5] 7[0] 0.4417 𝛽  0.0614** 4[2] 9[4] -0.0412 13[12] 13[10] 0.1026*** 𝛽   -0.0864 7[0] 6[1] -1.3639** 21[7] 5[2] 1.2775** 𝛽  -0.0462*** 8[5] 5[2] 0.0592** 11[1] 15[5] -0.1054*** 𝛽   0.2107** 3[1] 10[5] 0.1260 9[0] 17[4] 0.0847 𝑹 𝟐 adj. 97.53% 96.87% 56.41% This table presents regression estimates for the equally weighted portfolios of US green and conventional funds, as well as the difference between these two portfolios, obtained by from the Treynor and Mazuy (1966) extended to a conditional multifactor setting regressions with both S&P500 (Panel A) and KLD400 (Panel B) as benchmarks, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability (𝜶𝒑), systematic risk (𝜷𝒑), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination (𝑅 adj) . rm2, SMB2 HML2 and MOM2 refers to squared risk factors. The predetermined information variables are the short-term rate (ST) and the dividend yield (DY). The time-varying alphas and betas associated with the risk factors are omitted. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). N+ and Nindicate the number of the funds that have positive and negative estimates, respectively. Within brackets the number of funds whose estimates are statistically significant at a 5% significance level are presented. 42 5.3.4 The conditional five-factor version of the Treynor and Mazuy (1966) model Finally, table 11 presents the results for the Treynor and Mazuy (1966) extended to a conditional five-factor setting. Appendixes 15 and 16 detail the results for individual funds, for the full period time (2004-2019). Once again, in terms of selectivity green fund managers do not exhibit this ability, while conventional fund managers present a successful stock-picking ability. Additionally, conventional fund managers are better in terms of selectivity than green fund managers. However at the individual level, only few conventional funds present a positive and statistically significant alpha coefficient. The difference portfolio shows that green fund managers are better than conventional fund managers in timing the market, since the coefficient associated with the market squared risk factor is positive and statistically significant. In terms of investment styles, conventional fund managers do not show ability to time the book-to-market and the profitability factors correctly, since the squared coefficients associated with these risk factors are negative and statistically significant. In panel B, conventional fund managers are able to time the investment style. Focusing on differences between portfolios, green fund managers are better in timing the book-to-market and the profitability styles. Though they are worse in timing the investment style. 43 Table 11 – The conditional five-factor version of the Treynor and Mazuy (1966) model Panel A: S&P500 Portfolios Green (1) NN+ Conventional (2) NN+ Difference (1)-(2) 𝛼  -0.0011 9[0] 4[1] 0.0013* 8[0] 18[1] -0.0024** 𝛽  ∗  1.0239*** 0[0] 13[13] 0.9652*** 0[0] 26[26] 0.0587*** 𝛽  ∗   0.4785* 4[2] 9[2] -0.1176 14[2] 12[0] 0.5961** 𝛽  0.3125*** 0[0] 13[11] 0.3212*** 3[0] 23[20] -0.0086 𝛽   0.0700 5[1] 8[0] -0.6080 16[2] 10[0] 0.6780 𝛽  0.0422 5[3] 8[3] -0.0757*** 15[10] 11[9] 0.1179*** 𝛽   1.8657** 1[1] 12[5] -0.5991* 18[3] 8[1] 2.4649*** 𝛽  0.0096 5[0] 8[1] -0.0784** 16[10] 10[2] 0.0880** 𝛽   -0.1030 7[0] 6[0] -4.3414*** 20[7] 6[0] 4.2384*** 𝛽  -0.0268 8[1] 5[1] -0.1259*** 18[9] 8[2] 0.0991 𝛽   -4.4089* 10[5] 3[0] 2.1394 8[1] 18[1] -6.5483** 𝑹 𝟐 adj. 97.32% 98.65% 52.67% Panel B: MSCI KLD 400 Portfolios Green (1) NN+ Conventional (2) NN+ Difference (1)-(2) 𝛼  -0.0007 10[1] 3[1] 0.0019** 4[0] 21[2] -0.0026*** 𝛽  ∗  1.0170*** 0[0] 13[13] 0.9489*** 0[0] 26[26] 0.0680*** 𝛽  ∗   0.1993 4[2] 9[0] -0.5034** 19[5] 7[0] 0.7028** 𝛽  0.2843*** 1[1] 12[8] 0.2972*** 3[0] 23[18] -0.0129 𝛽   -0.0306 5[1] 8[0] -0.6005 14[2] 12[0] 0.5699 𝛽  0.1033*** 3[2] 10[5] -0.0190 15[10] 11[7] 0.1224*** 𝛽   0.7693 3[0] 10[3] -1.6549*** 21[9] 5[0] 2.4242*** 𝛽  0.0419 4[0] 9[1] -0.0494 14[7] 12[4] 0.0913** 𝛽   -0.0970 7[1] 6[0] -4.4077** 19[5] 7[0] 4.3107*** 𝛽  -0.0655 10[1] 3[1] -0.1581*** 18[9] 8[1] 0.0926 𝛽   -0.3326 7[0] 6[0] 6.0698*** 2[0] 24[7] -6.4024** 𝑹 𝟐 adj. 97.14% 97.20% 54.29% This table presents regression estimates for the equally weighted portfolios of US green and conventional funds, as well as the difference between these two portfolios, obtained by from the Treynor and Mazuy (1966) extended to a conditional multifactor setting regressions with both S&P500 (Panel A) and KLD400 (Panel B) as benchmarks, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability (𝜶𝒑), systematic risk (𝜷𝒑), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination (𝑅 adj) . rm2, SMB2 HML2, RMW2 and CMA2 refers to squared risk factors. The predetermined information variables are the short-term rate (ST) and the dividend yield (DY). The time-varying alphas and betas associated with the risk factors are omitted. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). N+ and Nindicate the number of the funds that have positive and negative estimates, respectively. Within brackets the number of funds whose estimates are statistically significant at a 5% significance level are presented. 44 5.4. Fund performance in different market states This study also analyzes the fund performance in different market states. This is performed by adding a dummy variable to distinguish periods of recessions and expansions periods. The dummy variable, which assumes a value of 1 in recession periods and 0 expansion periods, is added to the Carhart (1997) four-factor model and the Fama and French (2015) five-factor model. This analysis includes only 10 US green funds and the respective conventional funds, since the other funds were in existence mostly through only one market state. Table 12 reports the results for the Carhart (1997) four-factor model with the dummy variable. Appendixes 17 and 18 detail the results for individual funds, for the full period time (20042019). The alphas of the green and conventional portfolios and the coefficients of the dummy variables, that represent performance differentials in recession periods, are both statistically insignificant. This means that both green and conventional funds present a neutral performance in expansion periods, and there is no significant change of performance in recession periods. At the individual level, the majority of green and conventional funds also present a neutral performance in expansions, with no significant changes in recessions. An interesting result is how funds change systematic risk in periods of recession. While conventional funds significantly increase market risk in recessions, green funds tend to reduce their exposure to market risk in troubled times, since the coefficient for the dummy of the market risk is negative and even statistically significant in the case of the S&P500. This evidence is supported by the results at the individual fund level: in recessions periods several green funds significantly increase their exposure to the market, while several conventional funds significantly increase their level of systematic risk. Furthermore, in expansion periods conventional funds are more exposed to small stocks than green funds. Yet, in recession periods green funds not only significantly increase their exposure to small stocks, but they do so in a statistically different way than conventional funds. . 45 Regarding the HML factor, green funds are more exposed to value stocks than conventional funds in expansion periods. In troubled times, green funds reduce their exposure to value stocks only when the benchmark is the MSCI KLD 400. In the case of the S&P500, conventional funds Table 12Fund performance in different market states - Carhart (1997) four-factor model with a dummy variable Panel A: S&P500 Portfolios Green (1) NN+ Conventional (2) NN+ Difference (1)-(2) 𝛼  -0.0006 8[2] 2[0] -0.0000 12[1] 8[2] -0.0006 𝛼  -0.0016 6[1] 4[1] -0.0013 12[0] 6[2] -0.0004 𝛽  1.0243*** 0[0] 10[10] 0.9887*** 0[0] 20[20] 0.0356** 𝛽  -0.1416*** 9[6] 1[0] 0.0720*** 6[1] 14[6] -0.2136*** 𝛽  0.2514*** 1[0] 9[8] 0.3320*** 2[0] 18[17] -0.0806** 𝛽  ∗  0.4171*** 0[0] 10[6] 0.0413 6[1] 14[0] 0.3758** 𝛽  0.0187 6[2] 4[3] -0.1096*** 11[11] 9[7] 0.1282*** 𝛽  ∗  -0.0458 5[1] 5[1] -0.1866*** 19[10] 1[0] 0.1408*** 𝛽  -0.0857*** 8[4] 2[0] 0.0432** 9[1] 11[4] -0.1289*** 𝛽  ∗  -0.0456** 7[2] 3[1] -0.0268 10[4] 10[0] -0.0189 𝑹 𝟐 adj. 97.83% 98.13% 58.96% Panel B: MSCI KLD 400 Portfolios Green (1) NN+ Conventional (2) NN+ Difference (1)-(2) 𝛼  -0.0001 6[2] 4[1] 0.0006 7[1] 13[2] -0.0007 𝛼  -0.0024 6[1] 4[1] -0.0024 17[0] 3[1] 0.0000 𝛽  1.0179*** 0[0] 10[10] 0.9727*** 0[0] 20[20] 0.0453*** 𝛽  -0.0815 8[3] 2[1] 0.1445*** 0[0] 20[11] -0.2260*** 𝛽  0.2129*** 2[1] 8[7] 0.2999*** 3[1] 17[16] -0.0870** 𝛽  ∗  0.3212** 2[0] 8[5] -0.0815 13[2] 7[0] 0.4027** 𝛽  0.0838*** 2[1] 8[4] -0.0476 11[11] 9[8] 0.1314*** 𝛽  ∗  -0.1407** 8[3] 2[0] -0.2835*** 20[14] 0[0] 0.1428*** 𝛽  -0.0614*** 6[4] 4[4] 0.0646** 8[1] 12[5] -0.1260*** 𝛽  ∗  -0.0539** 8[3] 2[1] -0.0314 9[3] 11[0] -0.0224 𝑹 𝟐 adj. 97.36% 96.30% 58.87% This table presents regression estimates for the equally weighted portfolios of US green and conventional funds, as well as the difference between these two portfolios, obtained from the conditional four-factor model regressions with a dummy for both S&P500 (Panel A) and KLD400 (Panel B) as benchmarks, from February 2004 - September 2019. The dummy variable is added in order to distinguish recessions from expansions periods. It reports for both periods, estimates of performance (𝜶𝒑), the systematic risk (𝜷𝒑), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination (𝑅 adj. ). Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). N+ and Nindicate the number of the funds that have positive and negative estimates, respectively. Within brackets the number of funds whose estimates are statistically significant at a 5% significance level are presented. 46 tend to become even more exposed to growth stocks in recession periods, in a statistically different way than green funds. These results are supported at the individual level, since several conventional funds significantly increase their exposure to growth stocks in troubled times. Regarding the MOM factor, in expansion periods green funds are more exposed to companies with recent poor performance than conventional funds. In recession periods, green funds become even more exposed to companies with recent poor performance. Table 13 reports the estimates of performance for the conditional Fama and French (2015) five-factor model with a dummy variable. Appendixes 19 and 20 detail the results for individual funds, for the full period time (2004-2019). Once again, comparing the 𝑅𝑠 adj. of this model with the previous one, the 𝑅𝑠 adj. are slightly higher for conventional funds and slightly lower for green funds. In the case of the S&P500, in expansion periods the alpha coefficient of the green portfolio is negative and statistically significant at 10% level, so there is evidence that green funds tend to underperform the market, while conventional funds perform similarly to the market. Additionally, green funds perform worse than conventional funds in expansions periods, regardless of the benchmark used. Yet, the performance of conventional funds decreases significantly in troubled times, and in a statistically different way than green funds. These results are supported at the individual fund level. The majority of green and conventional funds present a neutral performance in expansion periods, but in recession periods green funds do not experience any significant decrease in performance, while several conventional funds do experience a worse performance in troubled times. With respect to systematic risk, in expansion periods green funds are more exposed to this risk than conventional funds. In troubled times, differently from the results in table 12, both green and conventional funds do not decrease or increase their exposure to market risk, since the coefficient for the dummy associated with market risk is statistically insignificant. As in the results in table 12, in expansion periods conventional funds are more exposed to small stocks than green funds. In troubled periods both green and conventional funds become even more exposed to small stocks, in the case of the S&P500. 53 6. Conclusion A growing number of investors introduce environmental screens into their investment decision process. In theory, environmental funds are subject to higher risks, since they limit the pool of investments, so it is important to study the issue of whether the inclusion of environmental screens punishes or improves the performance. This dissertation evaluates the performance of US green conventional funds, using unconditional and conditional models. Furthermore, analyzes fund managers’ abilities. Additionally, this study distinguishes the performance of these funds in expansion and recession periods. Varma and Nofsinger (2014) state that SRI attributes of companies make them less risky in recession periods, so it is important control for crisis periods. This study analyzed the performance of 13 US green funds and 26 matched US conventional funds relative to the market, using a conventional benchmark and socially responsible benchmark. Overall, both green and conventional funds present a neutral performance compared to the market. The results also indicate that green funds do not perform differently from conventional funds. So, focusing in the overall period the answer to the question “Did it pay to be a green investor?” is that for US domestic green funds, from 2004 to 2019 it does not pay, but it also doesn’t hurt. In terms of fund characteristics, green funds are more exposed to the socially responsible benchmark and conventional funds to the conventional benchmark. Regarding the size (SMB) factor, both funds seem to be more exposed to small stocks and there is no statistically significant difference in terms of this factor. In relation to the book-to-market (HML) factor, the results are not consistent among the models but, overall, green funds seem to be more exposed to value stocks and conventional funds to growth stocks. In terms of the difference portfolio, green funds are more exposed to value stocks than their conventional peers. Furthermore, the results also show that green funds are more exposed to companies with poor past performance, while conventional funds are more exposed to companies with a good past performance. Finally, the profitability and the investment factors from the five-factor model, in general, are only statistically significant for conventional funds, indicating that these funds are more exposed to companies with weak profitability and to high investments firms. In spite of the neutral performance of green funds in terms of these two additional risk factors, the results for the difference portfolio show that green funds are more exposed to companies with robust profitability and low investments compared to conventional funds. One possible explanation for the similar performance between green and conventional funds may be the possibility that SRI funds may not be much different from conventional funds, as 54 questioned by Leite et al. (2018). These authors argue that there are no strong boundaries between SRI and conventional funds. Investors who are willing to incorporate social concerns in their investment’s decisions cannot know more than the simple information contained in the prospectuses of the funds. It would be useful for investors to know more about the holdings of the funds that they invest in, more than the simple classification of the fund as being environmentally friendly. In fact, there is some evidence that SRI funds may not be much different from conventional funds (e.g., Utz and Wimmer, 2014), so in the same line or reasoning one might question whether green funds are ‘truly green’. The fact that there is no clear criteria to define what is a green fund further motivates this debate. As expected, the application of the conditional approach leads to an increase of the explanatory power of the models, as in Cortez et al. (2009). However, the performance remains neutral for both green and conventional funds. By applying conditional models, the results show evidence of time-varying betas, for both green and conventional funds, meaning that risk varies over time according to the public information variables. For the alphas, we cannot reject the null hypothesis of the time-varying alphas being equal to zero. Cortez et al. (2012) argue that this is not a surprise since green funds may lead to a more stable performance over time, since funds with social concerns could be more protected from a stock price drops, and consequently presenting more stability. This paper also analyzes timing and selectivity abilities of fund managers. Overall comparing the two types of funds, the results show a better selectivity ability for conventional fund managers, whereas green funds managers present better timing abilities. The combination of these abilities for each type of fund cancel out and result in a similar overall performance between the two type funds. When controlling the analysis for expansion and recession periods, there are some differences in terms of financial performance using the conditional Fama and French (2015) fivefactor model. Using this model, green funds underperform conventional funds in expansion periods. In recession periods conventional funds reduce their performance, while the performance of green funds remains unchanged. This means that in recession periods it is better to invest in green funds than in conventional funds since they maintain their performance, while in expansions periods is better to invest in conventional funds. However, using the conditional four-factor model, 55 there are not statistically significant differences between green and conventional funds, and both funds present a neutral performance in both periods. The results of this study are in line with the majority of the studies on the performance of green funds in the sense that green investors may expect no superior or inferior risk-adjusted returns by investing in green funds. The general evidence on the performance of conventional funds documents neutral or negative performance compared to the market, which is also in accordance with this study, since conventional funds also present a neutral performance compared to the market. In sum, investing in US green funds does not seems to punish the financial performance of investors compared to conventional funds. This conclusion has important implications for US investors, they can do well by choosing green funds, without sacrificing financial performance. Besides that, as mentioned by Silva and Cortez (2016) conventional investors can include environmental funds to diversify their portfolios. The results show some evidence that in expansion periods green funds underperform conventional funds, but in recession periods conventional funds reduce their performance while the performance of green funds do not increase or reduce, so invest in green funds is a good way to diversify and protect conventional investors in recessions periods. Regarding the limitations of this study, it is important to mention that since there is no clear definition of what a green fund is, this study makes the selection of green funds based on the “environmental” category defined by US SIF. As mentioned by Chang et al. (2012), the list of the US SIF may not be complete, we might have missed other green funds that are not listed in this source. Likewise, we did not consider funds that ceased to exist, so the results may be influenced by survivorship bias. Since selecting the green funds is one of the most important steps for this study, this is the main limitation of the study. Another limitation is associated to the socially responsible benchmark used (KLD400) that is oriented to socially responsible stocks in general and not green stocks specifically. The fact is that environmental equity indexes are more recent and using one of them would imply shortening the evaluation period in a considerable way. Since this study only has 13 US green funds, for further investigation, it would be interesting to extend the number of green funds and extend the analysis to a global scale, comparing the performance of green funds in different countries. In addition, since the Carhart (1997) four-factor model presents higher explanatory power for green funds and the Fama and 56 French (2015) for conventional funds, a suggestion for further investigation would be the use of a six-factor model, as in Fama and French (2018). 57 REFERENCES Adamo, R., Federico, D., & Notte, A. (2014). Performance and risk of green funds. Investment Management and Financial Innovations , (11, Iss. 1 (contin.)), 134-145. Adcock, C. J., Cortez, M. C., Armada, M. J. & F. Silva, F. (2012). Time varying betas and the unconditional distribution of asset returns . Quantitative Finance , 12(6), 951–967. Ambec, S., & Lanoie, P. (2008). Does it pay to be green? A systematic overview. The Academy of Management Perspectives , 45-62. Ang, W. R., Gregoriou, G. N., & Lean, H. H. (2014). Market-timing skills of socially responsible investment fund managers: The case of North America versus Europe. Journal of Asset Management , 15, 366–377. Areal, N., Cortez, M. C., & Silva, F. (2013). The conditional performance of US mutual funds over different market regimes: do different types of ethical screens matter?. Financial Markets and Portfolio Management , 27(4), 397-429. Aupperle, K. E., Carroll, A. B., & Hatfield, J. D. (1985). An empirical examination of the relationship between corporate social responsibility and profitability. Academy of Management Journal , 28(2), 446-463. Baum, C. F. (2006). An Introduction to Modern Econometrics Using Stata . Texas: Stata Press. Bernhardt, I., & Jung, B. S. (1979). The interpretation of least squares regression with interaction or polynomial terms. The Review of Economics and Statistics , 481-483. Boulatoff, C., & Boyer, C. M. (2009). Green recovery: how are environmental stocks doing?. The Journal of Wealth Management , 12(2), 9-20. Busch, T., & Lewandowski, S. (2018). Corporate Carbon and Financial Performance: A Meta‐analysis . Journal of Industrial Ecology , 22(4), 745-759. Carhart, M.M., (1997). On Persistence in Mutual Fund Performance. Journal of Finance , 52(1), pp.57–82. 58 Chang, C. E., Nelson, W. A., & Doug Witte, H. (2012). Do green mutual funds perform well?. Management Research Review , 35(8), 693-708. Christopherson, J.A., Ferson, W.E. & Glassman, D.A., (1998). Alphas on Economic Information: Another Look at the Persistence of Performance. Review of Financial Studies , 11(1), pp.111–142. Chung, H., Lee, H. H., & Tsai, P. C. (2012). Are green fund investors really socially responsible ?. Review of Pacific Basin Financial Markets and Policies , 15(04), 1250023. Climent, F. & Soriano, P., (2011). Green and Good? The Investment Performance of US 40 Environmental Mutual Funds. Journal of Business Ethics , 103(2), pp.275–287. Cordeiro, J. & Sarkis, J. (1997). “Environmental proactivism and firm performance: Evidence from security analyst earnings forecasts”. Business Strategy and the Environment , 6(1), 104-114. Cortez, M. C., Silva, F., & Areal, N. (2009). The performance of European socially responsible funds. Journal of Business Ethics , 87 (4), 573-588. Cortez, M. C., Silva, F., & Areal, N. (2012). Socially responsible investing in the global market: The performance of US and European funds. International Journal of Finance & Economics , 17(3), 254-271. Derwall, J., Guenster, N., Bauer, R. & Koedijk, K. (2005). “The Eco-Efficiency premium. Financial Analysts Journal , 61(2), 51-63. Elton, E. J., Gruber, M. J., Das, S., & Hlavka, M., (1993). Efficiency with costly information: A reinterpretation of evidence from managed portfolios. The Review of Financial Studies , 6(1), 122. Entreat, J., Guenther, E., & Hoppe, H. (2014). Making sense of conflicting empirical findings: A meta-analytic review of the relationship between corporate environmental and financial performance. European Management Journal , 32(5), 735-751. Eurosif, (2018). European SRI Study Fama, E. F., & French, K. R. (1993). Common risk factors in the returns on stocks and bonds. Journal of Financial Economics , 33 (1), 3-56. 59 Fama, E. F., & French, K. R. (2018). Choosing factors. Journal of Financial Economics , 128(2), 234-252. Fama, E.F. & French, K.R., (2015). A five-factor asset pricing model. Journal of Financial Economics , 116(1), pp.1–22. Feldman, S., Soyka, P., & Ameer, P. (1997). Does improving a firm’s environmental management system and environmental performance result in a higher stock price? Journal of Investing , 6(4), 87-97. Ferruz, L., Muñoz, F., & Vargas, M. (2010). Stock picking, market timing and style differences between socially responsible and conventional pension funds: Evidence from the United Kingdom. Business Ethics: A European Review , 19, 408–422. Ferson, W. E., & Schadt, R. W. (1996). Measuring fund strategy and performance in changing economic conditions. The Journal of Finance , 51 (2), 425-461.. Ferson, W. E., & Warther, V. A. (1996). Evaluating fund performance in a dynamic market. Financial Analysts Journal , 52(6), 20-28. Ferson, W. E., Sarkissian, S., & Simin, T. (2008). Asset pricing models with conditional betas and alphas: The effects of data snooping and spurious regression. Journal of Financial and Quantitative Analysis , 43(2), 331-353. Ferson, W., Sarkissian, S., & Simin, T. (2003). Is stock return predictability spurious. Journal of Investment Management , 1 (3), 1-10. Freeman, R. E.,(1984). Strategic management: A Stakeholder Approach . Pitman. Friedman, M., (1970). The Social Responsibility of Business Is to Increase its profits. New York Times Magazine , September 13, pp.112-6. Haan, M., Dam, L., & Scholtens, B. (2012). The drivers of the relationship between corporate environmental performance and stock market returns. Journal of Sustainable Finance & Investment , 2(3-4), 338-375.30-37. Hart, S.L. & Ahuja, G. (1996). Does it pay to be green? An empirical examination of the relationship between emission reduction and firm performance. Business Strategy and the Environment , 5(1), 30-37. 60 Ibikunle, G., & Steffen, T. (2017). European green mutual fund performance: A comparative analysis with their conventional and black peers. Journal of Business Ethics , 145 (2), 337-355. Jensen, M.C., (1968). The Performance of Mutual Funds in the Period 1945-1964. The Journal of Finance , 23(2), pp.389–416. Jo, H., Kim, H., & Park, K. (2015). Corporate environmental responsibility and firm performance in the financial services sector. Journal of business Ethics , 131(2), 257-284. Keefe, J.F., (2007). From Socially Responsible Investing to Sustainable Investing. Green Money Journal , 6(2). King, A. A. & Lenox, M. J. (2001). Does it really pay to be green? An empirical study of firm environmental and financial performance. Journal of Industrial Ecology , 5(1), 105-116. Klassen, R. D. & McLaughlin, C. P. (1996). The impact of environmental management on firm performance. Management Science , 42(8), 1199-1214. Konar, S. & Cohen, M. A. (2001). Does the market value environmental performance?. Review of Economics and Statistics , 83(2), 2814–2890. Leite, C., Cortez, M. C., Silva, F., & Adcock, C. (2018). The performance of socially responsible equity mutual funds: Evidence from Sweden. Business Ethics: A European Review , 27 (2), 108-126. Leite, P., & Cortez, M. C. (2014). Selectivity and timing abilities of international socially responsible funds. Applied Economics Letters , 21 (3), 185-188. Lesser, K., Rößle, F., & Walkshäusl, C. (2016). Socially responsible, green, and faith-based investment strategies: Screening activity matters!. Finance Research Letters , 16, 171-178. Lioui, A., & Sharma, Z. (2012). Environmental corporate social responsibility and financial performance: Disentangling direct and indirect effects. Ecological Economics , 78, 100-111. Mallett, J. E., & Michelson, S. (2010). Green investing: is it different from socially responsible investing ?. International Journal of Business , 15(4). Manrique, S., & Martí-Ballester, C. P. (2017). Analyzing the effect of corporate environmental performance on corporate financial performance in developed and developing countries . Sustainability , 9(11), 1957. 61 Marti‐Ballester, C. P. (2019a). The role of mutual funds in the sustainable energy sector. Business Strategy and the Environment , 28(6), 1107-1120. Martí-Ballester, C. P. (2019b). Do European renewable energy mutual funds foster the transition to a low-carbon economy?. Renewable Energy , 143, 1299-1309. McGuire, J., Sundgren, A. & Schneeweis, T. (1988), Corporate social responsibility and firm financial performance . Academy of Management Journal , Vol. 31 No. 4, pp. 854-72. Moskowitz, M.R., (1972) Choosing Socially Responsible Stocks. Business & Society Review , 1, pp.71–75. Muñoz, F., Vargas, M. & Marco, I., (2014). Environmental Mutual Funds: Financial Performance and Managerial Abilities . Journal of Business Ethics, 124(4), pp.551–569. Nakao, Y., Amano, A., Matsumura, K., Genba, K., & Nakano, M. (2007). Relationship between environmental performance and financial performance: an empirical analysis of Japanese corporations . Business Strategy and the Environment , 16(2), 106-118. Newey, W., & West, K. (1987). A simple, positive-definite, heteroskedascity and autocorrelation consistent covariance matrix . Econometrica, 55, 703-708. Nofsinger, J., & Varma, A. (2014). Socially responsible funds and market crises. Journal of Banking & Finance , 48, 180-193. Pekovic, S., Grolleau, G., & Mzoughi, N. (2018). Environmental investments: too much of a good thing?. International Journal of Production Economics , 197, 297-302. Puopolo, G. W., Teti, E., & Milani, V. (2015). Does the market reward for going green?. Journal of Management Development , 34(6), 729-742. Ramanathan, R. (2018). Understanding complexity: The curvilinear relationship between environmental performance and firm performance . Journal of Business Ethics, 149(2), 383-393. Reboredo, J. C., Quintela, M., & Otero, L. A. (2017). Do investors pay a premium for going green? Evidence from alternative energy mutual funds. Renewable and Sustainable Energy Reviews , 73, 512-520. Silva, F., & Cortez, M. C. (2016). The performance of US and European green funds in different market conditions. Journal of Cleaner Production , 135 , 558-566. 62 Solomon, R., & Hansen, K. (1985). It’s good business . New York: Atheneum . Stanwick, P. & Stanwick, S. (1998). The relationship between corporate social performance, and organizational size, financial performance, and environmental performance: An empirical examination. Journal of Business Ethics, 17(2), 195-204. Treynor, J., & Mazuy, K. (1966). Can mutual funds outguess the market? Harvard Business Review , 44, 131–136. Ullmann, A. A. (1985). Data in search of a theory: A critical examination of the relationships among social performance, social disclosure, and economic performance of US firms. Academy of Management Review , 10 (3), 540-557. US SIF, (2018). Report on Socially Responsible Investing Trends in the United States, Available at: The Forum for Sustainable and Responsible Investment. Wahba, H. (2008). Does the market value corporate environmental responsibility? An empirical examination. Corporate Social Responsibility and Environmental Management ,15(2), 8999. Walley, N., & Whitehead, B. (1994). It's not easy being green. Reader in Business and the Environment, 36, 81. White, M., (1995). The performance of environmental mutual funds in the United States and Germany: Is there economic hope for green investors? In J. Post, D. Collins, & M. Starik, eds. Research in corporate social performance and policy. Greenwich, CT: JAI Press , pp. 323–344. 69 Appendix 4 - Performance estimates using the unconditional the Fama and French (2015) five-factor model – Standard & Poor’s 500 - Conventional funds This table presents regression estimates for the US conventional funds, obtained by the regression of the five-factor model the S&P500 as benchmark, from February 2004 - September 2019. It reports estimates performance (𝛂𝐩), the systematic risk (𝛃𝐩), factor loadings associated to size (SMB), profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination (R adj.).Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). Standard & Poor`s 500 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 -0.0019** 0.0007 0.0005 -0.0024** -0.0008 -0.0003 0.0016** 0.0001 0.0028*** -0.0021 -0.0018* -0.0001 -0.0001 𝜷 𝒑 0.9266*** 1.0757*** 0.6853*** 0.9678*** 1.0123*** 0.9840*** 0.8586*** 0.9584*** 1.0430*** 1.0256*** 0.8597*** 1.0233*** 0.9314*** 𝜷 𝑺𝑴𝑩 0.1095*** 0.2518*** -0.0011 0.0890** 0.0653 0.1330*** 0.3113*** -0.0180 0.2726*** 0.5222*** 0.0834** 0.2802*** 0.1773*** 𝜷 𝑯𝑴𝑳 -0.0926* -0.3732*** -0.1892*** 0.1004** 0.1535** -0.2675*** 0.1785*** 0.0990*** -0.3633*** -0.5067*** -0.0293 0.0406** -0.0585* 𝜷 𝑹𝑴𝑾 -0.1557*** -0.3503*** 0.3390*** 0.0388 -0.0845 -0.1697* 0.0472 -0.0033 -0.3733*** -0.1044 -0.1663** 0.0131 -0.1286*** 𝜷 𝑪𝑴𝑨 -0.1348** -0.4612*** 0.2599** -0.2962*** -0.0736 -0.5094*** 0.1208* 0.0352 -0.4431*** -0.1616 -0.0179 -0.0250 -0.1161*** 𝑹 𝟐 adj. 0.9504 0.9216 0.8616 0.9726 0.9273 0.9227 0.9326 0.9673 0.9040 0.8414 0.9233 0.9946 0.9582 70 Appendix 4 - Performance estimates using the unconditional the Fama and French (2015) five-factor model – Standard & Poor’s 500 - Conventional fundscontinued This table presents regression estimates for the US conventional funds, obtained by the regression of the five-factor model the S&P500 as benchmark, from February 2004 - September 2019. It reports estimates performance (𝛂𝐩), the systematic risk (𝛃𝐩), factor loadings associated to size (SMB), profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination (R adj.).Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). Standard & Poor`s 500 x14 x15 x16 x17 x18 x19 x20 x21 x22 x23 x24 x25 x26 𝜶 𝒑 -0.0002 0.0008 0.0019* 0.0005 0.0008* 0.0001 -0.0009 -0.0006 0.0014 0.0019 -0.0004 -0.0017 0.0010 𝜷 𝒑 0.9639*** 1.0298*** 1.0420*** 0.9487*** 0.8587*** 1.0261*** 1.0605*** 0.9784*** 0.9389*** 0.9834*** 0.9724*** 0.8963*** 0.8734*** 𝜷 𝑺𝑴𝑩 -0.0433 0.1611*** 0.1337** 0.0525 0.0267 1.0060*** 0.4870*** 0.4505*** 0.8150*** 0.4003*** 0.5103*** 0.9504*** 0.3157*** 𝜷 𝑯𝑴𝑳 0.0711 -0.3867*** -0.3562*** 0.1070** 0.0802** 0.2542*** 0.0082 -0.2364*** -0.5014*** -0.1995*** -0.4202*** 0.3672*** -0.1270* 𝜷 𝑹𝑴𝑾 -0.0097 -0.2390*** -0.3570*** -0.0764 0.1139*** 0.1370*** -0.1271 -0.1559*** -0.3729*** -0.2051** -0.2111** 0.2859** 0.2094** 𝜷 𝑪𝑴𝑨 -0.0233 -0.1870** -0.2887*** 0.0485 0.1529*** -0.1160** -0.1418* -0.2722*** -0.2629** -0.1485* -0.1526* -0.1730 0.2589** 𝑹 𝟐 adj. 0.9377 0.9261 0.9059 0.9589 0.9484 0.9659 0.9067 0.9274 0.8855 0.9420 0.9156 0.9150 0.8718 71 Appendix 4 - Performance estimates using the unconditional the Fama and French (2015) five-factor model – MSCI KLD 400 - Conventional funds This table presents regression estimates for the US conventional funds, obtained by the regression of the five-factor model with the KLD400 as benchmark, from February 2004 - September 2019. It reports estimates performance (𝛂𝐩), the systematic risk (𝛃𝐩), factor loadings associated to size (SMB), profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination (R adj.). Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). MSCI KLD 400 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 -0.0018 0.0008 0.0006 -0.0023** -0.0007 -0.0005 0.0018** 0.0003 0.0031*** -0.0017 -0.0015 0.0002 0.0000 𝜷 𝒑 0.9211*** 1.0676*** 0.6580*** 0.9446*** 0.9860*** 0.9680*** 0.8602*** 0.9455*** 1.0279*** 0.9942*** 0.8498*** 1.0171*** 0.9352*** 𝜷 𝑺𝑴𝑩 0.0552 0.1886*** -0.0058 0.0554 0.0383 0.1061** 0.2637*** -0.0663* 0.2233*** 0.4793*** 0.0390 0.2249*** 0.1274*** 𝜷 𝑯𝑴𝑳 -0.0773 -0.3557*** -0.1749** 0.1333*** 0.1878** -0.2326*** 0.2098*** 0.1359*** -0.3145*** -0.4532*** -0.0012 0.0721** -0.0209 𝜷 𝑹𝑴𝑾 -0.1907*** -0.3928*** 0.3730*** 0.0649 -0.0350 -0.1121 0.0482 -0.0173 -0.3798*** -0.1234 -0.1891** -0.0101 -0.1212*** 𝜷 𝑪𝑴𝑨 -0.1581** -0.4859*** 0.2410* -0.3208*** -0.1004 -0.5429*** 0.0803 -0.0165 -0.5117*** -0.2377 -0.0638 -0.0751 -0.1682*** 𝑹 𝟐 adj. 0.9273 0.8992 0.8319 0.9614 0.9075 0.9264 0.9293 0.9411 0.8849 0.8077 0.8998 0.9786 0.9593 72 Appendix 4 - Performance estimates using the unconditional the Fama and French (2015) five-factor model – MSCI KLD 400 - Conventional funds - continued This table presents regression estimates for the US conventional funds, obtained by the regression of the five-factor model with the KLD400 as benchmark, from February 2004 - September 2019. It reports estimates performance (𝜶𝒑), the systematic risk (𝜷𝒑), factor loadings associated to size (SMB), profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination (𝑅 adj.). Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). MSCI KLD 400 x14 x15 x16 x17 x18 x19 x20 x21 x22 x23 x24 x25 x26 𝜶 𝒑 0.0001 0.0010 0.0022* 0.0009 0.0011 0.0004 -0.0004 -0.0004 0.0018 0.0018 -0.0005 -0.0015 0.0011 𝜷 𝒑 0.9532*** 1.0261*** 1.0313*** 0.9292*** 0.8456*** 1.0172*** 1.0361*** 0.9711*** 0.9174*** 0.9953*** 0.9803*** 0.8775*** 0.8648*** 𝜷 𝑺𝑴𝑩 -0.0900** 0.1086** 0.0829 0.0096 -0.0137 0.9555*** 0.4399*** 0.4019*** 0.7733*** 0.3423*** 0.4481*** 0.9275*** 0.2875*** 𝜷 𝑯𝑴𝑳 0.1150* -0.3424*** -0.3090*** 0.1533** 0.1206*** 0.3001*** 0.0608 -0.1931*** -0.4548*** -0.1977*** -0.4207*** 0.4014*** -0.1004 𝜷 𝑹𝑴𝑾 -0.0132 -0.2368*** -0.3600*** -0.0867 0.1081** 0.1352** -0.1405 -0.1567** -0.3848*** -0.2419*** -0.2444** 0.3268*** 0.2503** 𝜷 𝑪𝑴𝑨 -0.0849 -0.2487*** -0.3548*** -0.0169 0.0961* -0.1803*** -0.2162** -0.3329*** -0.3288** -0.1199 -0.1378 -0.1985 0.2468*** 𝑹 𝟐 adj. 0.9175 0.9175 0.8901 0.9273 0.9220 0.9564 0.8797 0.9158 0.8655 0.9348 0.9046 0.9078 0.8774 73 Appendix 5 - Performance estimates using the conditional the Carhart (1997) four-factor model - Standard & Poor`s 500Green funds This table presents regression estimates for the US green funds, obtained by the regression of the conditional four-factor model with the S&P500 as benchmark, from February 2004 - September 2019. It reports estimates of performance (𝛂𝐩), systematic risk (𝛃𝐩), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination (R adj.). The predetermined information variables are the short-term rate (ST) and the dividend yield (DY). Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). Standard & Poor`s 500 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 -0.0030*** -0.0026 0.0019 0.0009 0.0001 0.0016 -0.0006* -0.0009** 0.0017** -0.0021** -0.0018 0.0007 0.0004 𝜶 𝑺𝑻 -0.2144 -0.9359 0.6181 -0.5058*** -0.1253 -0.1633 -0.0856 -0.0253 -0.2755*** -0.1629 -0.0629 -0.0953 -0.0140 𝜶 𝑫𝒀 -0.0157*** -0.0949** 0.0309* 0.0072 0.0048 -0.0045 -0.0042** 0.0015 -0.0017 -0.0171** -0.0192** 0.0050 0.0340** 𝜷 𝒑 ∗ 𝒓𝒎 1.0448*** 1.0428*** 0.9895*** 1.0535*** 0.9996*** 0.8263*** 0.9874*** 0.9830*** 0.8646*** 1.1346*** 1.1855*** 0.9157*** 0.9673*** 𝜷 𝑺𝑻 ∗ 𝒓𝒎 14.1434** 56.6236 -3.6361 2.0242 -4.2290 -8.0170 0.0297 1.6447 -1.8722 6.4068 8.9394 12.4173 2.4021 𝜷 𝑫𝒀 ∗ 𝒓𝒎 -0.2144 2.3964 0.1611 -0.2650* 0.0436 -0.0536 0.0426 -0.1295** -0.3097*** -0.0799 0.1348 -0.0772 -0.5457 𝜷 𝑺𝑴𝑩 0.2403*** 0.2970 0.1734** 0.1513*** 0.3341*** 0.2838*** 0.1545*** 0.0316* 0.0732** 0.4397*** 0.6000*** 0.7676*** 0.5942*** 𝜷 𝑺𝑻 ∗ 𝑺𝑴𝑩 -2.1714 -14.7640 -28.0821 1.2797 -23.1274** -11.4838 -1.2740 -5.9193* -11.0732 -47.3430*** -41.1089** -5.7369 -55.6024*** 𝜷 𝑫𝒀 ∗ 𝑺𝑴𝑩 0.7589*** -0.3295 0.9331 -0.0332 0.3553* 0.2723* 0.0206 -0.1194 0.2354 0.3819 0.0512 -0.1055 -0.6790 𝜷 𝑯𝑴𝑳 0.1597*** -0.0649 -0.2341*** -0.0211 -0.0249 -0.1279*** 0.0013 -0.0785*** -0.0217 0.1057* 0.1514*** 0.0747 0.0804 𝜷 𝑺𝑻 ∗ 𝑯𝑴𝑳 10.9287 15.5859 -1.8999 -20.7353 -29.2846* -0.9120 2.5395 -18.5789*** 8.4506 4.3532 -15.0677 13.2927 -3.4939 𝜷 𝑫𝒀 ∗ 𝑯𝑴𝑳 0.2306 1.6293 1.0001 -0.2350 -1.0833*** -0.1032 -0.0469 -0.0436 0.3652** -0.5107* -0.5951* 0.2146 -0.1234 𝜷 𝑴𝑶𝑴 -0.2137*** -0.3002* 0.1818*** -0.1585*** -0.0370 -0.0772** -0.0095 -0.0110 0.0019 -0.1658*** -0.1504*** -0.0048 0.0253 𝜷 𝑺𝑻 ∗ 𝑴𝑶𝑴 -5.8963 30.4044 -19.4328 -9.9214 -8.2150 4.2562 -2.3556* -2.5780 -4.8177 -3.1059 1.9361 -9.4262 -14.9270 𝜷 𝑫𝒀 ∗ 𝑴𝑶𝑴 0.0433 0.5893 0.5648 -0.4963*** -0.5260*** -0.0345 -0.0836*** -0.1202* -0.1169 -0.2159 -0.3647 -0.1559 -0.4635 𝑹 𝟐 adj. 0.9438 0.8824 0.8927 0.9073 0.8854 0.9020 0.9888 0.9799 0.9242 0.9356 0.9304 0.9344 0.9290 74 Appendix 5 - Performance estimates using the conditional the Carhart (1997) four-factor model - MSCI KLD 400 – Green funds MSCI KLD 400 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 -0.0026*** -0.0018 0.0022 0.0011 0.0003 0.0017 -0.0005 -0.0008*** 0.0018*** -0.0018* -0.0015 0.0010 0.0006 𝜶 𝑺𝑻 -0.1287 -1.4357 0.4422 -0.5064*** -0.1116 -0.1675 -0.0738 -0.0235 -0.2607** -0.1306 -0.0306 -0.0302 -0.1227 𝜶 𝑫𝒀 -0.0168** -0.1165*** 0.0233 0.0057 0.0029 -0.0066 -0.0067** -0.0004 -0.0034 -0.0193** -0.0219* 0.0067 0.0265 𝜷 𝒑 ∗ 𝒓𝒎 1.0420*** 0.9670*** 0.9582*** 1.0539*** 1.0038*** 0.8229*** 0.9897*** 1.0002*** 0.8672*** 1.1365*** 1.1779*** 0.9170*** 0.9385*** 𝜷 𝑺𝑻 ∗ 𝒓𝒎 11.9583* 67.0220 4.1361 -3.1272 -8.0848 -11.1958 -1.7366 0.3456 -2.5054 5.0814 5.5884 11.0844 4.9609 𝜷 𝑫𝒀 ∗ 𝒓𝒎 -0.1290 2.2769 0.1294 -0.1514 0.1566 0.0406 0.1780** 0.0028 -0.2142* 0.0235 0.2788 -0.0650 -0.4382 𝜷 𝑺𝑴𝑩 0.2046*** 0.2898 0.1366** 0.1079** 0.2913*** 0.2512*** 0.1126*** -0.0179*** 0.0349 0.3945*** 0.5571*** 0.7385*** 0.5608*** 𝜷 𝑺𝑻 ∗ 𝑺𝑴𝑩 3.3161 -15.1738 -27.3071 8.7581 -16.9041 -5.1989 3.2401 -1.9940** -7.2806 -43.1009** -35.7007** -3.0322 -55.0396** 𝜷 𝑫𝒀 ∗ 𝑺𝑴𝑩 0.9226*** 0.7702 1.3725* 0.1164 0.4898** 0.3940** 0.1426 -0.0119 0.3520 0.5342 0.2099 0.0159 -0.0953 𝜷 𝑯𝑴𝑳 0.2084*** -0.1420 -0.2587*** 0.0344 0.0308 -0.0864* 0.0592*** -0.0184*** 0.0302 0.1744*** 0.2209*** 0.1061 0.0620 𝜷 𝑺𝑻 ∗ 𝑯𝑴𝑳 22.4485*** 39.8493 11.1571 -6.8993 -14.0125 9.9453 18.4251*** -1.9988** 21.7483** 22.6094* 3.5865 19.6057 7.6280 𝜷 𝑫𝒀 ∗ 𝑯𝑴𝑳 0.2194 0.5669 0.2403 -0.2640 -1.0837*** -0.1245 -0.0396 -0.0216 0.3909** -0.4540 -0.5849* 0.2592 -0.8022 𝜷 𝑴𝑶𝑴 -0.2027*** -0.3754** 0.1476** -0.1386*** -0.0146 -0.0634* 0.0138 0.0147*** 0.0203 -0.1402*** -0.1230*** 0.0015 -0.0038 𝜷 𝑺𝑻 ∗ 𝑴𝑶𝑴 -4.7106 53.5063 -5.7005 -8.1517 -5.5429 5.5960 1.3421 1.5067*** -1.4502 1.5466 6.1683 -6.3754 -2.4790 𝜷 𝑫𝒀 ∗ 𝑴𝑶𝑴 0.1198 0.6323 0.6549 -0.4067*** -0.4245** 0.0339 0.0249 -0.0053 -0.0296 -0.0973 -0.2388 -0.0485 -0.2693 𝑹 𝟐 adj. 0.9373 0.8755 0.8972 0.9105 0.8867 0.8943 0.9850 0.9985 0.9183 0.9299 0.9227 0.9296 0.9245 This table presents regression estimates for the US green funds, obtained by the regression of the conditional four-factor model with the KLD400 as benchmark, from February 2004 - September 2019. It reports estimates of performance ( 𝛂 𝐩 ), systematic risk ( 𝛃 𝐩), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination ( R  adj.). The predetermined information variables are the shortterm rate (ST) and the dividend yield (DY). Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 75 Appendix 6 - Performance estimates using the conditional the Carhart (1997) four-factor model - Standard & Poor`s 500 - Conventional funds Standard & Poor`s 500 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 -0.0028*** -0.0014 0.0008 -0.0038** -0.0004 -0.0026 0.0017* 0.0002 0.0010 -0.0041*** -0.0011 -0.0002 -0.0007 𝜶 𝑺𝑻 -0.2794 -0.0259 0.2853 0.0319 -0.7665 0.6410 0.0304 -0.1167 0.0954 0.5040* 0.0215 -0.0471 0.0676 𝜶 𝑫𝒀 -0.0081 -0.0147 0.0262 -0.0233* 0.0055 -0.0216 0.0033 0.0057** -0.0102 -0.0124 -0.0014 -0.0015 -0.0043 𝜷 𝒑 ∗ 𝒓𝒎 0.9191*** 1.1161*** 0.6910*** 1.0768*** 0.9893*** 1.0329*** 0.8475*** 0.9715*** 1.1076*** 1.1278*** 0.7951*** 1.0239*** 0.9247*** 𝜷 𝑺𝑻 ∗ 𝒓𝒎 -9.0352** -13.0708* 5.6132 -17.3707 5.5557 9.1555 1.4948 2.8388 -15.800*** -3.6153 -33.208*** 1.4117 -2.6475 𝜷 𝑫𝒀 ∗ 𝒓𝒎 0.0068 -0.0197 -0.2245 0.1993 -0.7071** -0.2410 -0.1292 0.1281** -0.1819 -0.0043 -0.2055* 0.0580 -0.1121 𝜷 𝑺𝑴𝑩 0.1261*** 0.3289*** -0.0642 0.1085** 0.0896* 0.2292*** 0.3048*** 0.0014 0.3612*** 0.5656*** 0.1615*** 0.2724*** 0.1826*** 𝜷 𝑺𝑻 ∗ 𝑺𝑴𝑩 -8.1371 7.3889 1.2583 -8.0581 -13.3244 -23.9042 -4.6853 -7.5756* -9.7117 -4.8711 10.1893 -3.0541 2.3860 𝜷 𝑫𝒀 ∗ 𝑺𝑴𝑩 0.0508 0.3997 0.5237 0.8990** -0.6282 0.5165 0.0553 -0.0559 0.1727 0.2722 0.2653 -0.0047 0.0666 𝜷 𝑯𝑴𝑳 -0.1535*** -0.5379*** 0.0906 0.1438** 0.1102 -0.3558*** 0.1847*** 0.1461*** -0.4662*** -0.3185*** -0.0306 0.0368*** -0.1324*** 𝜷 𝑺𝑻 ∗ 𝑯𝑴𝑳 16.4178** -23.7437 -31.1576 -46.9546* -15.7156 -36.9685 3.5251 33.8700*** -7.0597 41.0642*** 18.6695*** 2.9877 -12.6188** 𝜷 𝑫𝒀 ∗ 𝑯𝑴𝑳 -0.1461 -0.6270*** 0.7619 0.1198 0.6166 1.2231 0.0821 0.4850*** -0.5268** -0.2569 0.0261 -0.0895 -0.3512*** 𝜷 𝑴𝑶𝑴 -0.0672*** 0.0195 0.1433* 0.1503*** -0.0436 0.0689 -0.0801** -0.0276 0.0710 0.2396*** 0.0007 -0.0065 -0.0282 𝜷 𝑺𝑻 ∗ 𝑴𝑶𝑴 1.1095 -16.5517** -49.7161 -28.4914 0.0767 -5.3352 -4.0993 4.3903 -11.1136 -3.8704 -0.4925 0.3682 0.2121 𝜷 𝑫𝒀 ∗ 𝑴𝑶𝑴 -0.0159 -0.2351 0.4221 0.1697 -1.3197*** -0.5525 -0.0195 0.2540*** -0.2891* -0.0994 -0.0402 -0.0107 -0.2805*** 𝑹 𝟐 adj. 0.9571 0.9026 0.8109 0.9717 0.9284 0.8981 0.9334 0.9750 0.8879 0.8935 0.9493 0.9947 0.9637 This table presents regression estimates for the US conventional funds, obtained by the regression of the conditional four-factor model with the S&P500 as benchmark, from February 2004 - September 2019. It reports estimates of performance ( 𝛂 𝐩 ), systematic risk ( 𝛃 𝐩 ), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination ( R  adj.). The predetermined information variables are the short-term rate (ST) and the dividend yield (DY). Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 76 Appendix 6 - Performance estimates using the conditional the Carhart (1997) four-factor model - Standard & Poor`s 500 - Conventional funds - continued Standard & Poor`s 500 x14 x15 x16 x17 x18 x19 x20 x21 x22 x23 x24 x25 x26 𝜶 𝒑 0.0002 -0.0001 -0.0000 -0.0002 0.0011* 0.0010 -0.0019 -0.0011 -0.0005 0.0011 -0.0020 0.0003 0.0020 𝜶 𝑺𝑻 -0.0913 0.0809 -0.0334 -0.1725* 0.1525 -0.1534 0.2147 -0.0758 -0.2356 0.0559 0.1257 -1.3612*** -0.0410 𝜶 𝑫𝒀 -0.0042 -0.0031 -0.0060 0.0068* 0.0044 -0.0008 -0.0151* -0.0067 -0.0182** -0.0011 -0.0104 0.0225 0.0154 𝜷 𝒑 ∗ 𝒓𝒎 0.9257*** 1.0454*** 1.1147*** 0.9922*** 0.8721*** 0.9733*** 1.0329*** 0.9822*** 1.0290*** 0.9676*** 1.0266*** 0.9065*** 0.8436*** 𝜷 𝑺𝑻 ∗ 𝒓𝒎 -5.8773 -11.6739 -7.2462 8.3405*** 3.3048 -8.1849** -7.6083 -9.5430 -11.2201* -20.5871*** -8.5640 8.1982 -4.7654 𝜷 𝑫𝒀 ∗ 𝒓𝒎 0.1034 -0.0009 -0.1110 0.1011 -0.0371 0.1332 0.4051** 0.1822 -0.0796 0.0216 -0.0376 -0.3620 -0.5443 𝜷 𝑺𝑴𝑩 -0.0402 0.2006*** 0.2063*** 0.0904*** 0.0338 1.0017*** 0.5302*** 0.4661*** 0.8793*** 0.4179*** 0.5393*** 0.7873*** 0.2605*** 𝜷 𝑺𝑻 ∗ 𝑺𝑴𝑩 -0.3415 18.2563** -2.0518 -14.0350*** -16.9594*** 2.4172 1.5098 19.8527** 11.8902 -47.3320** 14.2532 30.5388 -29.6098 𝜷 𝑫𝒀 ∗ 𝑺𝑴𝑩 -0.1376 0.3801** 0.1551 0.1389 -0.0930 0.3697* 0.4475 0.1460 0.2507 -0.0807 0.1208 -2.3830 -1.0585 𝜷 𝑯𝑴𝑳 0.0387 -0.3942*** -0.3754*** 0.1505*** 0.1460*** 0.2145*** 0.0486 -0.2825*** -0.4338*** -0.2853*** -0.3995*** 0.3731*** -0.0856 𝜷 𝑺𝑻 ∗ 𝑯𝑴𝑳 16.1791** -19.5941** 15.6636 25.0040*** 16.0187*** 16.3189** 56.6089*** -12.5746 -0.0776 5.2747 -16.6300 -19.1498 -7.6029 𝜷 𝑫𝒀 ∗ 𝑯𝑴𝑳 0.2262 -0.6603*** -0.2294 0.1585 0.3456*** -0.0199 -0.6043** -0.7009*** -0.2741 -0.6390*** -0.7471*** -0.0483 -1.5540 𝜷 𝑴𝑶𝑴 0.0027 0.0916*** 0.0950** -0.0213 0.0216 -0.0494* -0.0188 0.0315 0.2591*** -0.0301 0.0570 0.0381 -0.0201 𝜷 𝑺𝑻 ∗ 𝑴𝑶𝑴 11.8718*** -6.4216 8.3073 9.2054*** 3.3685 -3.6540 6.5912 -7.1858 1.2725 -8.5964 -23.4834*** -18.3712 -29.4446 𝜷 𝑫𝒀 ∗ 𝑴𝑶𝑴 0.0200 -0.2451** -0.0760 0.3994*** 0.2786*** -0.0130 -0.1045 -0.2020* -0.2258 -0.1347 -0.4260** -1.2354 -0.4119 𝑹 𝟐 adj. 0.9475 0.9239 0.8941 0.9692 0.9520 0.9684 0.9359 0.9235 0.9031 0.9557 0.9212 0.9124 0.8518 This table presents regression estimates for the US conventional funds, obtained by the regression of the conditional four-factor model with the S&P500 as benchmark, from February 2004 - September 2019. It reports estimates of performance ( 𝛂 𝐩 ), systematic risk ( 𝛃 𝐩 ), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination ( R  adj.). The predetermined information variables are the short-term rate (ST) and the dividend yield (DY). Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 77 Appendix 6 - Performance estimates using the conditional the Carhart (1997) four-factor model - MSCI KLD 400 - Conventional funds MSCI KLD 400 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 -0.0027** -0.0014 0.0017 -0.0026 0.0003 -0.0024 0.0019* 0.0004 0.0013 -0.0038** -0.0010 0.0001 -0.0006 𝜶 𝑺𝑻 -0.2705 -0.0405 0.0197 -0.2851 -0.9328 0.4685 0.0513 -0.1133 0.0966 0.5139 0.0023 0.0047 0.0745 𝜶 𝑫𝒀 -0.0118 -0.0202 0.0186 -0.0325** 0.0026 -0.0297 0.0020 0.0027 -0.0142 -0.0163 -0.0051 -0.0042 -0.0067 𝜷 𝒑 ∗ 𝒓𝒎 0.9127*** 1.1148*** 0.6387*** 1.0286*** 0.9499*** 1.0009*** 0.8518*** 0.9613*** 1.0963*** 1.1132*** 0.7935*** 1.0167*** 0.9277*** 𝜷 𝑺𝑻 ∗ 𝒓𝒎 -9.5439 -10.2666 10.9926 -13.0689 7.3824 16.2305 -0.0774 -0.0828 -16.8464** -5.1685 -33.5374*** -0.7208 -3.9559 𝜷 𝑫𝒀 ∗ 𝒓𝒎 0.1507 0.1922 0.0657 0.4553 -0.5658 -0.3257 -0.0618 0.2541** 0.0175 0.2037 -0.0393 0.1987** 0.0210 𝜷 𝑺𝑴𝑩 0.0923* 0.2821*** -0.0751 0.0580 0.0664 0.1944*** 0.2707*** -0.0320 0.3149*** 0.5225*** 0.1248*** 0.2387*** 0.1414*** 𝜷 𝑺𝑻 ∗ 𝑺𝑴𝑩 -1.4181 13.6774 -0.2707 0.9405 -13.5818 -23.9552 -0.7183 -1.9262 -3.3188 0.6438 18.1602** 3.8388 6.7427 𝜷 𝑫𝒀 ∗ 𝑺𝑴𝑩 0.1861 0.5443 1.1903 2.0154*** 0.1655 0.9810 0.1719 0.0785 0.3204 0.4162 0.3837 0.1511 0.1811 𝜷 𝑯𝑴𝑳 -0.1235*** -0.4962*** 0.0593 0.1012* 0.0888 -0.3977*** 0.2356*** 0.1975*** -0.4045*** -0.2536*** 0.0040 0.0855*** -0.0780*** 𝜷 𝑺𝑻 ∗ 𝑯𝑴𝑳 23.8773** -13.2683 -18.6758 -23.8591 -1.3514 -19.9949 17.0881* 47.4672*** 9.0759 58.2217*** 27.9362*** 15.4962** 1.9822 𝜷 𝑫𝒀 ∗ 𝑯𝑴𝑳 -0.2307 -0.7180** 0.3257 -0.4551 -0.0165 0.2845 0.1281 0.4651*** -0.5692** -0.2981 -0.0486 -0.1327 -0.3497*** 𝜷 𝑴𝑶𝑴 -0.0602* 0.0320 0.1129 0.0932*** -0.0749 0.0241 -0.0614** -0.0095 0.0927 0.2642*** 0.0098 0.0077 -0.0068 𝜷 𝑺𝑻 ∗ 𝑴𝑶𝑴 1.5751 -14.2866 -36.6197 -5.2021 15.7615 11.0862 -0.9828 6.8084 -7.4699 0.2824 0.3789 1.9507 3.6738 𝜷 𝑫𝒀 ∗ 𝑴𝑶𝑴 0.0461 -0.1472 0.7160 0.6062 -1.0618** -0.5131 0.0662 0.3428** -0.1776 0.0219 0.0247 0.0757 -0.1804*** 𝑹 𝟐 adj. 0.9331 0.8780 0.7645 0.9630 0.9063 0.9050 0.9299 0.9556 0.8672 0.8740 0.9295 0.9802 0.9612 This table presents regression estimates for the US conventional funds, obtained by the regression of the conditional four-factor model with the KLD400 as benchmark, from February 2004 - September 2019. It reports estimates of performance ( 𝛂 𝐩 ), systematic risk ( 𝛃 𝐩 ), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination ( R  adj.). The predetermined information variables are the short-term rate (ST) and the dividend yield (DY). Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 78 Appendix 6 - Performance estimates using the conditional the Carhart (1997) four-factor model - MSCI KLD 400 - Conventional funds - continued MSCI KLD 400 x14 x15 x16 x17 x18 x19 x20 x21 x22 x23 x24 x25 x26 𝜶 𝒑 0.0004 0.0000 0.0002 0.0002 0.0014* 0.0012 -0.0018 -0.0009 -0.0003 0.0012 -0.0019 0.0011 0.0024 𝜶 𝑺𝑻 -0.0845 0.0666 -0.0355 -0.1358 0.1818 -0.1417 0.2198 -0.0792 -0.2356 0.1892 0.1783 -1.6145*** -0.1234 𝜶 𝑫𝒀 -0.0073 -0.0072 -0.0097 0.0045 0.0029 -0.0038 -0.0195* -0.0103 -0.0219** -0.0009 -0.0131 0.0157 0.0115 𝜷 𝒑 ∗ 𝒓𝒎 0.9185*** 1.0504*** 1.1091*** 0.9779*** 0.8629*** 0.9723*** 1.0255*** 0.9774*** 1.0149*** 0.9770*** 1.0290*** 0.8677*** 0.8259*** 𝜷 𝑺𝑻 ∗ 𝒓𝒎 -8.0698 -11.5550* -9.0936 5.5395 0.0381 -9.8636* -9.0099 -12.2613* -13.5893* -25.2739*** -12.2364 9.0379 -4.1120 𝜷 𝑫𝒀 ∗ 𝒓𝒎 0.2590 0.2098* 0.1045 0.2104** 0.0493 0.2579 0.5869*** 0.3706** 0.1261 0.1423 0.1307 -0.2261 -0.3528 𝜷 𝑺𝑴𝑩 -0.0768** 0.1501*** 0.1579** 0.0588* 0.0035 0.9618*** 0.4904*** 0.4248*** 0.8379*** 0.3852*** 0.4964*** 0.7715*** 0.2379*** 𝜷 𝑺𝑻 ∗ 𝑺𝑴𝑩 4.8958 22.9454*** 3.8716 -9.4585 -12.0277** 7.6409 6.6407 25.8770*** 18.3305 -46.0339** 19.6412 29.1258 -30.1201 𝜷 𝑫𝒀 ∗ 𝑺𝑴𝑩 -0.0182 0.4913** 0.2898 0.2826 0.0379 0.4988** 0.5724 0.2649 0.3856 0.0357 0.2601 -1.7062 -0.3660 𝜷 𝑯𝑴𝑳 0.0903** -0.3334*** -0.3114*** 0.2083*** 0.1957*** 0.2690*** 0.1061* -0.2287*** -0.3772*** -0.2589*** -0.3687*** 0.3343*** -0.0927 𝜷 𝑺𝑻 ∗ 𝑯𝑴𝑳 30.2862*** -2.4803 32.6984** 40.1043*** 28.7572*** 31.5000*** 73.2499*** 2.5258 14.7506 18.9596 -5.1621 -3.7796 3.2054 𝜷 𝑫𝒀 ∗ 𝑯𝑴𝑳 0.1963 -0.6882*** -0.2800 0.1704 0.3507*** -0.0124 -0.6267* -0.7537*** -0.3423 -0.6558** -0.8443*** -0.7865 -1.9956 𝜷 𝑴𝑶𝑴 0.0230 0.1168*** 0.1204** -0.0000 0.0395** -0.0287 0.0047 0.0540 0.2804*** -0.0197 0.0679 -0.0049 -0.0376 𝜷 𝑺𝑻 ∗ 𝑴𝑶𝑴 14.8138*** -2.0886 12.2224 12.6564*** 5.8730** -0.4403 10.3261* -4.2636 4.3139 -4.4162 -22.8600** -1.9377 -17.7145 𝜷 𝑫𝒀 ∗ 𝑴𝑶𝑴 0.1181 -0.1222 0.0463 0.4991*** 0.3614*** 0.0874 0.0114 -0.0940 -0.1198 0.0304 -0.3322 -1.0263 -0.1132 𝑹 𝟐 adj. 0.9329 0.9201 0.8843 0.9437 0.9337 0.9611 0.9212 0.9165 0.8893 0.9523 0.9140 0.8967 0.8533 This table presents regression estimates for the US conventional funds, obtained by the regression of the conditional four-factor model with the KLD400 as benchmark, from February 2004 - September 2019. It reports estimates of performance ( 𝛂 𝐩 ), systematic risk ( 𝛃 𝐩 ), factor loadings associated to size (SMB), book-tomarket (HML) and momentum (MOM) factors and the adjusted coefficient of determination ( R  adj.). The predetermined information variables are the short-term rate (ST) and the dividend yield (DY). Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 85 Appendix 9 - Selectivity and timing abilitiesUnconditional Carhart (1997) four-factorStandard & Poor’s 500 – Green funds This table presents regression estimates for the US green funds, obtained by from the Treynor and Mazuy (1966) extended to a multifactor setting regressions with S&P500 as benchmark, from February 2004 - September 2019. It reports estimates of performance (𝜶𝒑), systematic risk (𝜷𝒑), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination (𝑅 adj) . rm2, SMB2 HML2 and MOM2 refers to squared risk factors. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). Standard & Poor`s 500 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 -0.0010 0.0010 0.0013 -0.0006 -0.0005 0.0023* -0.0005 -0.0014** 0.0020** -0.0011 -0.0014 -0.0030* -0.0004 𝜷 𝒑 ∗ 𝒓𝒎 0.9969*** 1.1698*** 0.9694*** 1.0485*** 1.0451*** 0.8442*** 0.9916*** 0.9804*** 0.8423*** 1.1224*** 1.2003*** 0.9073*** 0.9744*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 -0.1248 -5.1291** 0.5290 0.3521 0.2512 -0.7367*** -0.1101 0.2497 -0.5330** 0.0043 0.0417 0.8456* 1.0654* 𝜷 𝑺𝑴𝑩 0.2229*** 0.2347 0.1412*** 0.1879*** 0.3664*** 0.2714*** 0.1574*** 0.0427** 0.0456 0.4250*** 0.6035*** 0.7602*** 0.5421*** 𝜷 𝑺𝑴𝑩𝟐 -1.9507 4.1058 -0.9388 -0.1872 -2.2415 0.1466 -0.1449 0.2910 0.0428 -2.1250 -1.4721 1.4590 -2.8453** 𝜷 𝑯𝑴𝑳 0.1545*** -0.0135 -0.2607*** -0.0329 -0.0368 -0.1275*** -0.0062 -0.0484** -0.0217 0.1181** 0.2062*** 0.0391 0.0867* 𝜷 𝑯𝑴𝑳𝟐 0.5725 -4.1025 -1.0001 1.8011** 1.9191** 0.5204 0.2257 -0.0796 0.3118 0.6718 -0.7481 1.0490* 2.1631** 𝜷 𝑴𝑶𝑴 -0.1688*** -0.2422** 0.1366** -0.1350*** -0.0862 -0.1349*** -0.0055 -0.0158 0.0352 -0.1486*** -0.1421*** 0.0830* 0.0367 𝜷 𝑴𝑶𝑴𝟐 -0.0124 2.7996 1.1744 0.3211*** 0.3122** -0.0451 0.0503 0.0084 0.2890*** 0.4297*** 0.7830*** 0.2540* -0.7474 𝑹 𝟐 adj. 0.9302 0.8881 0.8958 0.8998 0.8712 0.9001 0.9885 0.9768 0.9175 0.9234 0.9280 0.9374 0.9266 86 Appendix 9 - Selectivity and timing abilitiesUnconditional Carhart (1997) four-factorStandard & Poor’s 500 – Green funds MSCI KLD 400 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 0.0001 0.0007 0.0013 0.0001 0.0003 0.0030** 0.0003 -0.0008*** 0.0027*** -0.0001 -0.0004 -0.0024 -0.0003 𝜷 𝒑 ∗ 𝒓𝒎 0.9988*** 1.1629*** 0.9570*** 1.0618*** 1.0557*** 0.8467*** 0.9954*** 1.0025*** 0.8487*** 1.1257*** 1.2027*** 0.9030*** 0.9544*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 -0.5857 -5.2218* -0.0662 -0.0504 -0.1463 -1.1387*** -0.4850 -0.0474 -0.9058** -0.4236 -0.4014 0.7841 0.4470 𝜷 𝑺𝑴𝑩 0.1696*** 0.1696 0.0900* 0.1279** 0.3078*** 0.2287*** 0.1057*** -0.0182*** 0.0003 0.3667*** 0.5418*** 0.7209*** 0.4932*** 𝜷 𝑺𝑴𝑩𝟐 -2.3346 3.8620 -0.7994 -0.4105 -2.5176* -0.0128 -0.4040 0.0433 -0.1614 -2.4229 -1.7929 1.2536 -2.7468** 𝜷 𝑯𝑴𝑳 0.1698*** 0.0198 -0.2545*** -0.0064 -0.0007 -0.1072** 0.0281 -0.0134** 0.0069 0.1568** 0.2477*** 0.0582 0.0895* 𝜷 𝑯𝑴𝑳𝟐 0.5339 -3.5000 -0.6634 1.8709** 1.9920** 0.5220 0.2393 0.0451 0.3597 0.6847 -0.7477 1.0365* 2.5073*** 𝜷 𝑴𝑶𝑴 -0.1627*** -0.2480*** 0.1146** -0.1159*** -0.0616 -0.1241*** 0.0151 0.0114** 0.0532* -0.1254*** -0.1176*** 0.0784* 0.0123 𝜷 𝑴𝑶𝑴𝟐 0.0115 3.1405 1.4756** 0.3488*** 0.3554*** -0.0066 0.0928 0.0444*** 0.3365*** 0.4766*** 0.8309*** 0.1840 -0.4352 𝑹 𝟐 adj. 0.9237 0.8745 0.9031 0.9066 0.8746 0.8930 0.9825 0.9984 0.9154 0.9188 0.9231 0.9311 0.9245 This table presents regression estimates for the US green funds, obtained by from the Treynor and Mazuy (1966) extended to a multifactor setting regressions with KLD400 as benchmark, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability ( 𝜶 𝒑), systematic risk ( 𝜷 𝒑), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination ( 𝑅  adj) . rm2, SMB2 HML2 and MOM2 refers t o squared risk factors. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 87 Appendix 10 - Selectivity and timing abilitiesUnconditional Carhart (1997) four-factorStandard & Poor’s 500 – Conventional funds Standard & Poor`s 500 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 -0.0015 0.0022 0.0014 -0.0016 0.0009 -0.0006 0.0023** 0.0004 0.0031** -0.0020 -0.0007 0.0001 -0.0003 𝜷 𝒑 ∗ 𝒓𝒎 0.9296*** 1.1366*** 0.6698*** 1.0465*** 0.9997*** 1.0265*** 0.8244*** 0.9478*** 1.1180*** 1.1215*** 0.8515*** 1.0223*** 0.9371*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 -0.3830 -0.8411** -0.9330 -0.1745 0.1432 0.3812 -0.4940** 0.0312 -0.5014 -0.8283* -0.8884* -0.0101 -0.1608 𝜷 𝑺𝑴𝑩 0.1310*** 0.3304*** -0.0833 0.0944** 0.1004*** 0.2173*** 0.3048*** -0.0144 0.3578*** 0.5461*** 0.1070** 0.2810*** 0.2027*** 𝜷 𝑺𝑴𝑩𝟐 -1.8240* -2.5822** -0.9087 -0.7236 -3.5048* 0.2131 0.5715 0.5140 -1.6815 -1.9960 -0.1097 -0.4946 -1.1604 𝜷 𝑯𝑴𝑳 -0.1790*** -0.4869*** -0.0142 0.0875 0.1183** -0.4238*** 0.1581*** 0.0983*** -0.4650*** -0.3958*** -0.0539 0.0254* -0.1333*** 𝜷 𝑯𝑴𝑳𝟐 0.2849 -0.9709 0.2191 -0.1486 -0.0893 -1.8047* 1.0943** -0.4825 -0.3930 1.7415 -0.4748 0.2351 0.5164 𝜷 𝑴𝑶𝑴 -0.0564** 0.0472 0.0317 0.1074*** -0.0323 0.0453 -0.0892*** -0.0264 0.0501 0.2624*** -0.0166 -0.0107 -0.0512** 𝜷 𝑴𝑶𝑴𝟐 0.2261** 0.1997 1.0912* -0.2393 0.2802 0.4210 -0.2235** -0.1315* 0.0328 0.0395 0.1930* -0.0027 0.1890*** 𝑹 𝟐 adj. 0.9528 0.9042 0.8157 0.9679 0.9294 0.8922 0.9381 0.9674 0.8837 0.8831 0.9246 0.9947 0.9610 This table presents regression estimates for the US conventional funds, obtained by from the Treynor and Mazuy (1966) extended to a multifactor setting regressions with S&P500 as benchmark, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability ( 𝜶 𝒑), systematic risk ( 𝜷 𝒑), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination ( 𝑅  adj) . rm2, SMB2 HML2 and MOM2 refers to squared risk factors. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 88 Appendix 10 - Selectivity and timing abilitiesUnconditional Carhart (1997) four-factorStandard & Poor’s 500 – Conventional funds - continued Standard & Poor`s 500 x14 x15 x16 x17 x18 x19 x20 x21 x22 x23 x24 x25 x26 𝜶 𝒑 -0.0002 -0.0003 0.0005 0.0007 0.0010 0.0009 0.0013 -0.0006 0.0004 0.0018 -0.0022 -0.0026 0.0008 𝜷 𝒑 ∗ 𝒓𝒎 0.9475*** 1.0918*** 1.1187*** 0.9640*** 0.8487*** 0.9958*** 1.0584*** 1.0252*** 1.0616*** 0.9963*** 1.0550*** 0.9383*** 0.8455*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 -0.4018** -0.1307 -0.0083 0.2319 -0.1725 -0.7668** -1.5893*** 0.2783 -0.3986 -0.5191 0.5489* 0.7185 1.0166 𝜷 𝑺𝑴𝑩 -0.0623* 0.2079*** 0.1949*** 0.0965*** 0.0084 0.9621*** 0.5323*** 0.4969*** 0.8847*** 0.4421*** 0.5681*** 0.8884*** 0.2460*** 𝜷 𝑺𝑴𝑩𝟐 0.6317 -0.4257 -0.9261 -2.3655*** 1.4014 0.6000 -3.4860** -2.4549** -2.0304 -0.5712 -1.7209 1.3422 -2.3000 𝜷 𝑯𝑴𝑳 0.0456 -0.4054*** -0.4009*** 0.1145*** 0.1443*** 0.1856*** -0.0884 -0.2979*** -0.4329*** -0.2897*** -0.3942*** 0.3853*** -0.0489 𝜷 𝑯𝑴𝑳𝟐 -0.7272* 0.6690 -0.2186 1.5651*** 0.4956 0.1744 2.6638 0.2855 0.4627 1.0474 0.9668 -1.9139 2.4990 𝜷 𝑴𝑶𝑴 -0.0025 0.0614** 0.0780* -0.0312* 0.0187 -0.0246 -0.0574 0.0215 0.2449*** -0.0811* 0.0966** 0.0368 -0.0041 𝜷 𝑴𝑶𝑴𝟐 0.3929*** -0.0186 0.2659* -0.2925*** -0.3320*** 0.3111*** 0.1418 0.0168 0.2257* -0.1319 -0.0808 0.6451 -0.5539 𝑹 𝟐 adj. 0.9446 0.9202 0.8910 0.9651 0.9519 0.9676 0.9175 0.9208 0.9005 0.9404 0.9174 0.9065 0.8542 This table presents regression estimates for the US conventional funds, obtained by from the Treynor and Mazuy (1966) extended to a multifactor setting regressions with S&P500 as benchmark, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability ( 𝜶 𝒑), systematic risk ( 𝜷 𝒑), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination ( 𝑅  adj) . rm2, SMB2 HML2 and MOM2 refers to squared risk factors. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 89 This table presents regression estimates for the US conventional funds, obtained by from the Treynor and Mazuy (1966) extended to a multifactor setting regressions with KLD400 as benchmark, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability (𝜶𝒑), systematic risk (𝜷𝒑), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination (𝑅 adj) . rm2, SMB2 HML2 and MOM2 refers to squared risk factors. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). Appendix 10 - Selectivity and timing abilities -- Unconditional Carhart (1997) four-factor - MSCI KLD 400 - Conventional funds MSCI KLD 400 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 -0.0004 0.0034* 0.0018 -0.0012 0.0015 -0.0004 0.0029*** 0.0014 0.0043*** -0.0008 0.0002 0.0011 0.0005 𝜷 𝒑 ∗ 𝒓𝒎 0.9232*** 1.1324*** 0.6454*** 1.0134*** 0.9725*** 1.0085*** 0.8310*** 0.9366*** 1.1094*** 1.1061*** 0.8461*** 1.0178*** 0.9433*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 -0.8095* -1.3408** -1.2602 -0.9618** -0.5970 -0.3707 -0.7975*** -0.4018 -1.0346** -1.3010** -1.3206** -0.4429 -0.5511* 𝜷 𝑺𝑴𝑩 0.0876* 0.2754*** -0.1061 0.0381 0.0551 0.1672*** 0.2614*** -0.0571 0.3061*** 0.4989*** 0.0690 0.2332*** 0.1523*** 𝜷 𝑺𝑴𝑩𝟐 -2.2872** -3.1014* -1.1300 -0.8438 -3.4972* 0.2558 0.3004 0.2997 -1.9758 -2.2900 -0.2130 -0.6673 -1.4007** 𝜷 𝑯𝑴𝑳 -0.1718*** -0.4770*** -0.0104 0.0902* 0.1274** -0.4191*** 0.1833*** 0.1213*** -0.4278*** -0.3577*** -0.0405 0.0432 -0.1013*** 𝜷 𝑯𝑴𝑳𝟐 0.1712 -1.0835 0.5657 0.2785 0.0354 -1.4361 1.1027* -0.5697 -0.4340 1.6164 -0.5725 0.1144 0.5624 𝜷 𝑴𝑶𝑴 -0.0572 0.0476 0.0178 0.0709** -0.0594 0.0150 -0.0740*** -0.0178 0.0684 0.2778*** -0.0137 -0.0046 -0.0309 𝜷 𝑴𝑶𝑴𝟐 0.2465* 0.2297 1.2402** 0.1358 0.6736 0.7801 -0.1873** -0.1039 0.0926 0.0943 0.2197 0.0101 0.2344*** 𝑹 𝟐 adj. 0.9284 0.8819 0.7871 0.9543 0.9113 0.8941 0.9348 0.9420 0.8645 0.8550 0.8550 0.8550 0.9593 90 Appendix 10 - Selectivity and timing abilities -- Unconditional Carhart (1997) four-factor - MSCI KLD 400 - Conventional funds - continued MSCI KLD 400 x14 x15 x16 x17 x18 x19 x20 x21 x22 x23 x24 x25 x26 𝜶 𝒑 0.0009 0.0007 0.0017 0.0020* 0.0019** 0.0020 0.0024 0.0004 0.0016 0.0024 -0.0014 -0.0018 0.0010 𝜷 𝒑 ∗ 𝒓𝒎 0.9408*** 1.0945*** 1.1119*** 0.9453*** 0.8419*** 0.9909*** 1.0439*** 1.0195*** 1.0513*** 1.0075*** 1.0591*** 0.9107*** 0.8322*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 -0.8583** -0.6108* -0.5909** -0.3660 -0.5851** -1.3281*** -2.0468** -0.1975 -0.9708** -0.8970 0.1750 -0.0399 0.6079 𝜷 𝑺𝑴𝑩 -0.1064*** 0.1512*** 0.1411** 0.0558* -0.0311 0.9145*** 0.4890*** 0.4472*** 0.8362*** 0.3930*** 0.5118*** 0.8470*** 0.2067*** 𝜷 𝑺𝑴𝑩𝟐 0.3824 -0.7115 -1.2314 -2.6475*** 1.1721 0.3506 -3.7415** -2.7426** -2.3133 -0.5351 -2.1843 1.1297 -2.4079 𝜷 𝑯𝑴𝑳 0.0771* -0.3686*** -0.3645*** 0.1443*** 0.1723*** 0.2178*** -0.0522 -0.2639*** -0.3985*** -0.2842*** -0.3888*** 0.3903*** -0.0394 𝜷 𝑯𝑴𝑳𝟐 -0.7548 0.7025 -0.2124 1.5232*** 0.4709 0.1880 2.5307 0.2769 0.4392 1.1041 0.9438 -1.6317 2.7578* 𝜷 𝑴𝑶𝑴 0.0132 0.0837** 0.0979* -0.0180 0.0328* -0.0077 -0.0443 0.0403 0.2618*** -0.0943* 0.0974** 0.0031 -0.0186 𝜷 𝑴𝑶𝑴𝟐 0.4440*** 0.0351 0.3263** -0.2367** -0.2881*** 0.3776*** 0.2032 0.0632 0.2880* -0.1945 -0.0883 0.9669 -0.3094 𝑹 𝟐 adj. 0.9245 0.9134 0.8749 0.9335 0.9310 0.9583 0.8898 0.9074 0.8845 0.9345 0.9056 0.8951 0.8568 This table presents regression estimates for the US conventional funds, obtained by from the Treynor and Mazuy (1966) extended to a multifactor setting regressions with KLD400 as benchmark, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability ( 𝜶 𝒑), systematic risk ( 𝜷 𝒑), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination ( 𝑅  adj) . rm2, SMB2 HML2 and MOM2 refers to squared risk factors. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 91 Appendix 11 - Selectivity and timing abilitiesUnconditional Fama and French (2015) - Standard & Poor`s 500 - Green funds This table presents regression estimates for the US green funds, obtained by from the Treynor and Mazuy (1966) extended to a multifactor setting regressions with S&P500 as benchmark, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability (𝜶𝒑), systematic risk (𝜷𝒑), factor loadings associated to size (SMB), book-to-market (HML), profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination (𝑅 adj) . rm2, SMB2 HML2, RMW2 and CMA2 refers to squared risk factors. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). Standard & Poor`s 500 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 -0.0015 0.0032 0.0025 -0.0016 -0.0005 0.0017 -0.0004 -0.0009 0.0029** -0.0021 -0.0028 -0.0035* -0.0011 𝜷 𝒑 ∗ 𝒓𝒎 1.0640*** 1.2612*** 0.9136*** 1.1150*** 1.0631*** 0.8995*** 0.9965*** 0.9835*** 0.8629*** 1.1996*** 1.2850*** 0.9261*** 0.9544*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 0.2494 -3.3349** 0.3999 1.0039 0.8537 -0.4609 -0.0357 0.3260* -0.4311** 0.7218 1.1037 0.5998* 0.8266 𝜷 𝑺𝑴𝑩 0.1896*** 0.3283** 0.0669 0.1837*** 0.3106*** 0.2641*** 0.1556*** 0.0316* 0.0723* 0.4054*** 0.6017*** 0.8199*** 0.5816*** 𝜷 𝑺𝑴𝑩𝟐 -0.8350 -1.1006 -0.1400 1.1299 -0.3189 0.5689 0.0037 0.4505 0.4904 -0.1433 1.4216 1.3655 -2.6949* 𝜷 𝑯𝑴𝑳 0.3128*** 0.1456 -0.2318*** 0.1732** 0.0719 -0.0800 0.0133 -0.0337 -0.0378 0.2596*** 0.3822*** -0.0546 0.0952 𝜷 𝑯𝑴𝑳𝟐 2.5434*** -3.6652 -0.7701 3.9615*** 3.0886*** 1.3821** 0.4972*** 0.1818 0.7335 2.6061** 1.4494 0.5720 0.9086 𝜷 𝑹𝑴𝑾 -0.0435 -0.2169 -0.1299 0.0983 -0.1649 0.0269 0.0119 -0.0098 0.1555*** 0.0413 0.1085 0.2175** 0.1240 𝜷 𝑹𝑴𝑾𝟐 2.5675 -7.3109 -5.6340 0.7806 -2.7734 -3.7195 -0.0859 -1.5721 -0.8003 -1.5766 -4.3304 3.0109 0.8737 𝜷 𝑪𝑴𝑨 -0.0261 -0.0615 -0.3778*** -0.2213* -0.0727 0.1602* -0.0260 0.0168 0.0785 -0.0147 -0.0944 0.1546 -0.0861 𝜷 𝑪𝑴𝑨𝟐 -14.0215*** 12.7959 6.3578* -11.4756** -5.8564 -2.4399 -2.3402*** -2.2241 -6.8131** -9.5579* -8.5476 0.8241 3.7669 𝑹 𝟐 adj. 0.9178 0.8708 0.9047 0.8859 0.8638 0.8876 0.9884 0.9768 0.9199 0.9041 0.9037 0.9412 0.9264 92 Appendix 11 - Selectivity and timing abilitiesUnconditional Fama and French (2015) - MSCI KLD 400 – Green funds MSCI KLD 400 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 -0.0012 -0.0001 0.0017 -0.0016 -0.0002 0.0020 -0.0000 -0.0007*** 0.0032*** -0.0018 -0.0025 -0.0035* -0.0019 𝜷 𝒑 ∗ 𝒓𝒎 1.0606*** 1.2580*** 0.9117*** 1.1209*** 1.0650*** 0.8980*** 0.9933*** 0.9978*** 0.8626*** 1.1947*** 1.2777*** 0.9262*** 0.9477*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 -0.1233 -3.0893* -0.0065 0.7071 0.5880 -0.8095 -0.4083 -0.0081 -0.7939*** 0.4601 0.9113 0.4247 0.3812 𝜷 𝑺𝑴𝑩 0.1396* 0.2624 0.0293 0.1303* 0.2642*** 0.2264*** 0.1158*** -0.0144*** 0.0379 0.3558*** 0.5483*** 0.7776*** 0.5441*** 𝜷 𝑺𝑴𝑩𝟐 -1.2533 -1.5815 -0.0248 0.7409 -0.7478 0.3972 -0.2886 0.1003 0.2765 -0.6136 0.8664 1.0341 -2.6048* 𝜷 𝑯𝑴𝑳 0.3216*** 0.1414 -0.2177*** 0.1899*** 0.0989 -0.0638 0.0397 -0.0132** -0.0153 0.2938*** 0.4204*** -0.0607 0.1102* 𝜷 𝑯𝑴𝑳𝟐 2.1474*** -2.7302 -0.6424 3.6369*** 2.7891*** 1.1269 0.2394 -0.0018 0.5372 2.2202* 0.9932 0.2955 1.0748 𝜷 𝑹𝑴𝑾 -0.0559 -0.2259 -0.0906 0.1065 -0.1527 0.0316 0.0218 0.0100 0.1661** 0.0505 0.1161 0.1953* 0.1667 𝜷 𝑹𝑴𝑾𝟐 3.0708 0.1722 -3.5903 1.5080 -2.1152 -3.3505 0.4862 -0.7166** -0.2694 -1.0223 -3.8284 3.6010 3.0205 𝜷 𝑪𝑴𝑨 -0.0554 -0.0820 -0.3916*** -0.2586** -0.1195 0.1284 -0.0698* -0.0196* 0.0419 -0.0723 -0.1591 0.1844 -0.1043 𝜷 𝑪𝑴𝑨𝟐 -9.8052** 17.0091 8.6309** -7.8435 -2.7618 0.5912 0.6216 0.5325 -4.2668 -6.0059 -4.7348 3.8881 5.9802 𝑹 𝟐 adj. 0.9100 0.8573 0.9132 0.8953 0.8686 0.8810 0.9826 0.9984 0.9152 0.9016 0.9015 0.9361 0.9294 This table presents regression estimates for the US green funds, obtained by from the Treynor and Mazuy (1966) extended to a multifactor setting regressions with KLD400 as benchmark, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability ( 𝜶 𝒑), systematic risk ( 𝜷 𝒑), factor loadings associated to size (SMB), book-to-market (HML), profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination ( 𝑅  adj) . rm2, SMB2 HML2, RMW2 and CMA2 refers to squared risk factors. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of sign ificance of 1% (***), 5% (**) and 10% (*). 93 Appendix 12 - Selectivity and timing abilities - Unconditional Fama and French (2015) - Standard & Poor`s 500 - Conventional funds Standard & Poor`s 500 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 -0.0023* 0.0026 0.0024 -0.0011 0.0002 -0.0002 0.0013 0.0004 0.0049*** 0.0020 -0.0010 -0.0000 -0.0006 𝜷 𝒑 ∗ 𝒓𝒎 0.9311*** 1.0574*** 0.6969*** 0.9704*** 1.0091*** 0.9784*** 0.8667*** 0.9563*** 1.0268*** 1.0064*** 0.8371*** 1.0241*** 0.9458*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 0.0674 -0.4278 -1.4230* -0.5397 0.3261 0.6619 -0.5939*** -0.0416 -0.0994 -1.2014* -0.5460 0.0365 0.2239 𝜷 𝑺𝑴𝑩 0.1078*** 0.2759*** -0.0088 0.0908** 0.0737* 0.1455*** 0.3050*** -0.0206 0.2864*** 0.5574*** 0.0922** 0.2865*** 0.1615*** 𝜷 𝑺𝑴𝑩𝟐 -0.1351 -1.0648 0.4560 0.2825 -3.6499** 0.4303 0.1440 0.2181 -1.0828 -2.8931 1.0774 -0.5581 0.0430 𝜷 𝑯𝑴𝑳 -0.0887 -0.3966*** -0.1969** 0.0977* 0.1643** -0.2565*** 0.1787*** 0.0992*** -0.3574*** -0.5185*** -0.0644 0.0408** -0.0384 𝜷 𝑯𝑴𝑳𝟐 0.7879 -1.2410* -1.1314 -0.0277 -0.2893 -1.0854 1.5610*** -0.3760 -0.3635 0.2579 -0.8743 0.2957 1.4378*** 𝜷 𝑹𝑴𝑾 -0.1449*** -0.3259*** 0.3403*** 0.0550 -0.0909 -0.1590* 0.0488 -0.0091 -0.3121*** -0.0317 -0.1353** 0.0256 -0.1392*** 𝜷 𝑹𝑴𝑾𝟐 -1.2521 -2.4739 -4.2944 -4.7570 0.2661 -3.7224 2.3368 0.6031 -6.1977* -3.2660 -6.0020*** -1.0638* 0.9688 𝜷 𝑪𝑴𝑨 -0.1487** -0.4148*** 0.3158** -0.2685*** -0.0718 -0.5081*** 0.1077 0.0439 -0.4301*** -0.1128 0.0256 -0.0299 -0.1412*** 𝜷 𝑪𝑴𝑨𝟐 0.2092 4.5559 4.5044 0.2133 3.6113 1.6193 -2.2166 -1.0653 1.7708 -0.0576 7.6090** 1.0342 -5.7883** 𝑹 𝟐 a dj. 0.9496 0.9218 0.8596 0.9713 0.9295 0.9200 0.9344 0.9666 0.9040 0.8454 0.9288 0.9947 0.9600 This table presents regression estimates for the US conventional funds, obtained by from the Treynor and Mazuy (1966) extended to a multifactor setting regressions with S&P500 as benchmark, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability ( 𝜶 𝒑), systematic risk ( 𝜷 𝒑), factor loadings associated to size (SMB), book-to-market (HML), profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination ( 𝑅  adj) . rm2, SMB2 HML2, RMW2 and CMA2 refers to squared risk factors. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 94 Appendix 12 - Selectivity and timing abilities - Unconditional Fama and French (2015) - Standard & Poor`s 500 - Conventional funds - continued Standard & Poor`s 500 x14 x15 x16 x17 x18 x19 x20 x21 x22 x23 x24 x25 x26 𝜶 𝒑 0.0001 0.0014 0.0034** 0.0010 0.0015 -0.0007 0.0017 0.0004 0.0042** 0.0029* 0.0004 -0.0025 0.0022 𝜷 𝒑 ∗ 𝒓𝒎 0.9564*** 1.0224*** 1.0281*** 0.9625*** 0.8535*** 1.0233*** 1.0652*** 0.9763*** 0.9104*** 0.9817*** 0.9663*** 0.9027*** 0.8675*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 -0.0088 -0.0009 0.5131 0.0596 -0.6385** -0.4413 -1.1534** 0.5748* -0.3667 -0.1983 0.5549* 0.6186 0.3295 𝜷 𝑺𝑴𝑩 -0.0676* 0.1649*** 0.1215** 0.0748*** 0.0329 1.0016*** 0.5017*** 0.4707*** 0.8448*** 0.3893*** 0.5353*** 0.9656*** 0.3230*** 𝜷 𝑺𝑴𝑩𝟐 2.0113 -0.2413 0.4274 -3.0332*** -0.0391 1.2101 -2.6716* -2.3786** -1.5032 -0.1213 -2.1711 1.8757 -2.3351* 𝜷 𝑯𝑴𝑳 0.0766 -0.3821*** -0.3281*** 0.1212*** 0.0639* 0.2368*** 0.0172 -0.2134*** -0.5102*** -0.1916*** -0.4209*** 0.3674*** -0.1312* 𝜷 𝑯𝑴𝑳𝟐 -0.2751 0.3093 -0.1471 1.6598*** -0.0213 0.4828 3.6563* 0.4388 -1.2114 1.6368* 0.0278 -2.9465 0.6857 𝜷 𝑹𝑴𝑾 -0.0046 -0.1970*** -0.2931*** -0.0567 0.1156*** 0.1413*** -0.0510 -0.0968* -0.2905*** -0.1713* -0.1591* 0.2908** 0.1999** 𝜷 𝑹𝑴𝑾𝟐 -2.9446 -4.9064 -9.4632*** 1.0809 1.1129 -0.9997 -3.7095 -5.7317*** -7.9050* -7.0785** -7.4015** -3.8366 -6.9154** 𝜷 𝑪𝑴𝑨 -0.0116 -0.1924** -0.2970*** 0.0196 0.1774*** -0.1132* -0.1473 -0.3019*** -0.2308* -0.1436* -0.1579 -0.1460 0.2716** 𝜷 𝑪𝑴𝑨𝟐 -2.3817 2.2139 -1.1632 -1.7879 0.6532 3.7672 -4.6240 1.9750 6.4339 -0.1181 5.9388 6.9844 3.7801 𝑹 𝟐 adj. 0.9387 0.9257 0.9096 0.9627 0.9497 0.9662 0.9164 0.9296 0.8880 0.9449 0.9177 0.9136 0.8718 This table presents regression estimates for the US conventional funds, obtained by from the Treynor and Mazuy (1966) extended to a multifactor setting regressions with S&P500 as benchmark, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability ( 𝜶 𝒑), systematic risk ( 𝜷 𝒑), factor loadings associated to size (SMB), book-to-market (HML), profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination ( 𝑅  adj) . rm2, SMB2 HML2, RMW2 and CMA2 refers to squared risk factors. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 101 Appendix 14 - Selectivity and timing abilities - Conditional Carhart (1997) four-factor - MSCI KLD 400Conventional funds MSCI KLD 400 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 0.0001 0.0037 0.0018 -0.0010 0.0024 -0.0055** 0.0035*** 0.0018* 0.0052*** -0.0004 0.0008 0.0017** 0.0011 𝜷 𝒑 ∗ 𝒓𝒎 0.9024*** 1.0923*** 0.6806*** 1.0499*** 0.9706*** 0.9935*** 0.8513*** 0.9537*** 1.0705*** 1.1042*** 0.7866*** 1.0104*** 0.9218*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 -0.6383* -1.4306*** -1.3510 -0.9300 -1.5879* 1.2716 -1.2126*** -0.6212** -0.7050* -1.1934** -0.4326 -0.5592** -0.3920 𝜷 𝑺𝑴𝑩 0.0962** 0.2776*** -0.0805 0.0587 0.0723 0.2445*** 0.2561*** -0.0429 0.3103*** 0.5222*** 0.1256*** 0.2332*** 0.1445*** 𝜷 𝑺𝑴𝑩𝟐 -1.7933* -1.5525 -5.5437** 0.7792 -2.5440 2.1322 0.4566 0.3131 -1.4832 -2.6400 -1.2338 -0.3398 -1.4878** 𝜷 𝑯𝑴𝑳 -0.0929** -0.4453*** 0.0452 0.1562** 0.1057 -0.3754*** 0.2304*** 0.2086*** -0.3636*** -0.2312*** 0.0264 0.0979*** -0.0650** 𝜷 𝑯𝑴𝑳𝟐 -1.9054** -3.8645*** 4.1898*** -1.7970 -0.0162 -3.5951*** 0.5197 -1.2044* -3.6572** -1.1819 -1.4399 -1.0335** -0.7981 𝜷 𝑴𝑶𝑴 -0.0494 0.0300 0.1203 0.1056** -0.0769 -0.0061 -0.0906*** -0.0187 0.0785 0.2623*** 0.0231 0.0035 -0.0063 𝜷 𝑴𝑶𝑴𝟐 0.3019** 0.2540 0.4574 0.1369 0.7735 2.0985*** -0.2306 -0.0079 -0.0389 0.2495 0.2984* 0.0554 0.1153* 𝑹 𝟐 adj. 0.9376 0.8864 0.7724 0.9621 0.9081 0.9112 0.9357 0.9565 0.8717 0.8767 0.9323 0.9812 0.9622 This table presents regression estimates for the US conventional funds, obtained by from the Treynor and Mazuy (1966) extended to a conditional multifactor setting regressions wit KLD400 as benchmark, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability ( 𝜶 𝒑), systematic risk ( 𝜷 𝒑), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination ( 𝑅  adj) . rm2, SMB2 HML2 and MOM2 refers to squared risk factors. The predetermined information variables are the short-term rate (ST) and the dividend yield (DY). The time-varying alphas and betas associated with the risk factors are omitted. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 102 Appendix 14 - Selectivity and timing abilities - Conditional Carhart (1997) four-factor - MSCI KLD 400Conventional funds - continued MSCI KLD 400 x14 x15 x16 x17 x18 x19 x20 x21 x22 x23 x24 x25 x26 𝜶 𝒑 0.0005 0.0013 0.0027* 0.0024** 0.0015* 0.0021 0.0036** 0.0012 0.0019 0.0022 -0.0008 0.0022 0.0023 𝜷 𝒑 ∗ 𝒓𝒎 0.9205*** 1.0429*** 1.0883*** 0.9729*** 0.8655*** 0.9756*** 1.0107*** 0.9643*** 1.0078*** 0.9679*** 1.0233*** 0.8814*** 0.8272*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 -0.2197 -0.4324 -0.2395 -0.6405* -0.5657* -1.3595*** -1.8702*** 0.0108 -0.2699 -0.0665 0.1515 -1.0742 0.4686 𝜷 𝑺𝑴𝑩 -0.0825** 0.1446*** 0.1513** 0.0679** -0.0071 0.9354*** 0.4915*** 0.4357*** 0.8519*** 0.3811*** 0.5083*** 0.7613*** 0.2268*** 𝜷 𝑺𝑴𝑩𝟐 -0.3990 -0.2246 -0.9818 -2.1523** 1.0103 0.7657 -3.8131*** -2.2436* -2.9330 -0.0211 -1.7338 2.0540 -4.3453** 𝜷 𝑯𝑴𝑳 0.0972** -0.3212*** -0.2724*** 0.2157*** 0.1888*** 0.2747*** 0.1364** -0.2051*** -0.3607*** -0.2526*** -0.3550*** 0.3661*** -0.1562* 𝜷 𝑯𝑴𝑳𝟐 -0.5662 -1.0702 -3.8006*** 0.0890 0.6405 -0.3783 -1.4233 -1.8562** -0.7951 -0.8714 -0.6477 -1.4741 4.3588** 𝜷 𝑴𝑶𝑴 0.0469 0.1105*** 0.1318** -0.0164 0.0258 -0.0165 -0.0137 0.0568 0.2839*** -0.0390 0.0737 0.0003 -0.0339 𝜷 𝑴𝑶𝑴𝟐 0.4872*** -0.0070 0.3258 -0.1457* -0.1550* 0.4879*** 0.1054 0.1025 0.1586 -0.3690* 0.0643 -0.1739 -0.4564 𝑹 𝟐 adj. 0.9367 0.9193 0.8901 0.9469 0.9352 0.9650 0.9317 0.9180 0.8891 0.9528 0.9123 0.8910 0.8576 This table presents regression estimates for the US conventional funds, obtained by from the Treynor and Mazuy (1966) extended to a conditional multifactor setting regressions wit KLD400 as benchmark, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability ( 𝜶 𝒑), systematic risk ( 𝜷 𝒑), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination ( 𝑅  adj) . rm2, SMB2 HML2 and MOM2 refers to squared risk factors. The predetermined information variables are the short-term rate (ST) and the dividend yield (DY). The time-varying alphas and betas associated with the risk factors are omitted. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 103 Appendix 15 - Selectivity and timing abilities - Conditional Fama and French (2015) - Standard & Poor`s 500 - Green funds This table presents regression estimates for the US green funds, obtained by from the Treynor and Mazuy (1966) extended to a conditional multifactor setting regressions with S&P500 as benchmark, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability (𝜶𝒑), systematic risk (𝜷𝒑), factor loadings associated to size (SMB), book-to-market (HML), profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination (𝑅 adj) . rm2, SMB2, HML2, RMW2 and CMA2 refers to squared risk factors. The predetermined information variables are the short-term rate (ST) and the dividend yield (DY). The time-varying alphas and betas associated with the risk factors are omitted. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). Standard & Poor`s 500 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 -0.0016 0.0076 0.0037 -0.0009 -0.0004 0.0023* -0.0004 -0.0010 0.0025** -0.0034* -0.0041 -0.0026 -0.0003 𝜷 𝒑 ∗ 𝒓𝒎 1.0832*** 1.2204*** 0.8902*** 1.1059*** 1.0120*** 0.8388*** 0.9921*** 0.9886*** 0.8860*** 1.1678*** 1.2259*** 0.9423*** 0.9610*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 0.2681 -7.0778*** 0.3561 0.6954 0.7491* -0.0614 -0.0322 0.2969* -0.4579** 1.3356** 1.9634** 0.6621* 0.5334 𝜷 𝑺𝑴𝑩 0.2113*** 0.4194** 0.0867 0.1594** 0.3330*** 0.3030*** 0.1509*** 0.0192 0.0877** 0.4612*** 0.6590*** 0.8279*** 0.6144*** 𝜷 𝑺𝑴𝑩𝟐 -1.2104 8.3942 -0.9498 1.1847 0.4618 -0.0074 -0.1658 0.2225 0.1135 0.1934 1.2031 0.7916 -3.2661** 𝜷 𝑯𝑴𝑳 0.3227*** 0.3867 -0.2755*** 0.1096 -0.0350 -0.1502*** 0.0077 -0.0777*** -0.0366 0.2336*** 0.2491*** 0.0151 0.0711 𝜷 𝑯𝑴𝑳𝟐 2.9481*** -7.7950** 0.2797 4.5385** 1.7677 1.3196 0.6222** 0.9578** 1.6367* 1.1696 1.7634 1.3481 3.0392*** 𝜷 𝑹𝑴𝑾 -0.1077 -0.1423 -0.1890* 0.0299 -0.1745 0.0082 0.0050 -0.0323 0.1030* 0.0161 0.1367 0.2068** 0.0650 𝜷 𝑹𝑴𝑾𝟐 -2.2461 9.4675 -3.2577 2.0595 1.6034 -0.6795 -0.1931 -0.7692 -0.3436 0.8771 -1.1852 1.0161 0.5929 𝜷 𝑪𝑴𝑨 -0.0129 -0.3975 -0.3777** -0.1513 0.0296 0.2644*** -0.0054 0.0131 0.0817 -0.0744 -0.0850 0.1090 -0.1634 𝜷 𝑪𝑴𝑨𝟐 -12.1090*** 2.4138 5.9122 -13.6394** -5.1045 -3.7253 -2.1753** -3.2854** -6.8427** -8.2047 -9.5663 -0.3470 0.9685 𝑹 𝟐 adj. 0.9291 0.8940 0.9062 0.8913 0.8775 0.8993 0.9885 0.9799 0.9302 0.9230 0.9185 0.9449 0.9343 104 Appendix 15 - Selectivity and timing abilities - Conditional Fama and French (2015) - MSCI KLD 400 – Green funds MSCI KLD 400 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 -0.0010 -0.0005 0.0026 -0.0006 -0.0001 0.0026* -0.0000 -0.0008*** 0.0029*** -0.0031 -0.0037 -0.0020 -0.0012 𝜷 𝒑 ∗ 𝒓𝒎 1.0713*** 1.2366*** 0.8853*** 1.0995*** 1.0089*** 0.8298*** 0.9867*** 0.9979*** 0.8808*** 1.1575*** 1.2062*** 0.9374*** 0.9529*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 0.0209 -6.5151** 0.2314 0.4578 0.4481 -0.3629 -0.3952 0.0040 -0.7265** 1.1419 1.8051* 0.2409 0.0993 𝜷 𝑺𝑴𝑩 0.1809*** 0.2905 0.0488 0.1253* 0.3035*** 0.2784*** 0.1233*** -0.0144** 0.0649* 0.4318*** 0.6286*** 0.7889*** 0.5787*** 𝜷 𝑺𝑴𝑩𝟐 -1.3262 7.6619 -1.2369 0.9639 0.3283 -0.0043 -0.1769 0.0960 0.0309 0.0316 1.0585 1.0949 -3.4837*** 𝜷 𝑯𝑴𝑳 0.3800*** 0.3425 -0.2741*** 0.1738** 0.0235 -0.1019* 0.0655*** -0.0204** 0.0159 0.3064*** 0.3274*** 0.0231 0.0735 𝜷 𝑯𝑴𝑳𝟐 1.7664** -6.3352 0.4225 3.3936** 0.6995 0.3555 -0.3986 0.0650 0.8223 -0.2007 0.1612 0.0507 3.1305*** 𝜷 𝑹𝑴𝑾 -0.0902 -0.1419 -0.1717 0.0698 -0.1420 0.0353 0.0395 0.0084 0.1348** 0.0448 0.1636 0.1788* 0.0929 𝜷 𝑹𝑴𝑾𝟐 -2.2088 18.9418 -0.6702 2.5076 1.7334 -0.4267 -0.1986 -0.6808** -0.4238 0.9061 -1.2505 0.8385 3.2718 𝜷 𝑪𝑴𝑨 -0.0530 -0.4651 -0.3968*** -0.1957* -0.0075 0.2295** -0.0389 -0.0219 0.0518 -0.1245 -0.1406 0.1444 -0.1840 𝜷 𝑪𝑴𝑨𝟐 -7.7922* 11.3598 8.8711 -9.7288 -1.1208 -0.2361 1.8713 0.5042 -3.4807 -3.4270 -4.4239 3.7365 3.9255 𝑹 𝟐 adj. 0.9233 0.8887 0.9162 0.8973 0.8799 0.8924 0.9854 0.9984 0.9253 0.9201 0.9136 0.9394 0.9360 This table presents regression estimates for the US green funds, obtained by from the Treynor and Mazuy (1966) extended to a conditional multifactor setting regressions with KLD400 as benchmark, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability ( 𝜶 𝒑), systematic risk ( 𝜷 𝒑), factor loadings associated to size (SMB), book-to-market (HML), profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination ( 𝑅  adj) . rm2, SMB2, HML2, RMW2 and CMA2 refers to squared risk factors. The predetermined information variables are the short-term rate (ST) and the dividend yield (DY). The time-varying alphas and betas associated with the risk factors are omitt ed. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 105 Appendix 16 - Selectivity and timing abilities - Conditional Fama and French (2015) - Standard & Poor`s 500 - Conventional funds Standard & Poor`s 500 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 -0.0026* 0.0023 0.0047 -0.0000 -0.0001 -0.0024 0.0017 0.0006 0.0041** 0.0001 -0.0011 0.0000 -0.0004 𝜷 𝒑 ∗ 𝒓𝒎 0.9108*** 1.0840*** 0.6779*** 1.0000*** 0.9828*** 0.9727*** 0.8723*** 0.9656*** 1.0535*** 1.0821*** 0.7880*** 1.0208*** 0.9307*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 0.4397 -0.6545* -1.6924 -1.0502** 1.0033 1.8306* -0.7618** -0.2016 -0.2027 -1.3738* 0.1294 0.0256 0.4869* 𝜷 𝑺𝑴𝑩 0.1120*** 0.2512*** -0.0390 0.1255* 0.0513 0.1507** 0.3036*** -0.0163 0.2683*** 0.5514*** 0.1692*** 0.2828*** 0.1791*** 𝜷 𝑺𝑴𝑩𝟐 -0.2342 -0.1806 -3.0458* 2.6454 -2.9494* 1.8214 0.1704 0.0988 -0.5029 -2.8166* 0.1322 -0.3820 -0.1408 𝜷 𝑯𝑴𝑳 -0.0458 -0.3489*** -0.2881*** 0.0256 0.2167** -0.1079 0.1898*** 0.1507*** -0.3069*** -0.3433*** -0.0714 0.0610*** -0.0657* 𝜷 𝑯𝑴𝑳𝟐 -0.3869 -2.6983*** 0.5543 -1.8389 -0.4199 -1.6806 1.4848** -0.6948 -2.4434* -1.4005 -1.0531 -0.2048 1.5418* 𝜷 𝑹𝑴𝑾 -0.145*** -0.3642*** 0.2972 0.1205 -0.1940** -0.2012* 0.0062 -0.0254 -0.3293*** -0.0114 -0.0580 0.0187 -0.1420*** 𝜷 𝑹𝑴𝑾𝟐 0.3655 -4.8574 -8.4934 -3.1341 -1.7929 -4.9819 1.9839 -0.7040 -6.3153** -7.8907* -2.0905 -1.2399 2.3193 𝜷 𝑪𝑴𝑨 -0.1685** -0.4801*** 0.4475** -0.0627 -0.1046 -0.5830*** 0.1122 0.0618 -0.5236*** -0.2634* 0.0975 -0.0422* -0.1512*** 𝜷 𝑪𝑴𝑨𝟐 -0.2221 6.2779* 7.6394 -3.2797 4.1173 -0.1946 -1.4322 0.9639 4.1099 7.0552 4.7239* 1.6128* -7.6430** 𝑹 𝟐 adj. 0.9554 0.9281 0.8688 0.9713 0.9261 0.9279 0.9331 0.9724 0.9169 0.8792 0.9520 0.9948 0.9611 This table presents regression estimates for the US conventional funds, obtained by from the Treynor and Mazuy (1966) extended to a conditional multifactor setting regressions with S&P500 as benchmark, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability ( 𝜶 𝒑), systematic risk ( 𝜷 𝒑), factor loadings associated to size (SMB), book-to-market (HML), profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination ( 𝑅  adj) . rm2, SMB2, HML2, RMW2 and CMA2 refers to squared risk factors. The predetermined information variables are the short-term rate (ST) and the dividend yield (DY). The time-varying alphas and betas associated with the risk factors are omitted. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 106 Appendix 16 - Selectivity and timing abilities - Conditional Fama and French (2015) - Standard & Poor`s 500 - Conventional funds - continued Standard & Poor`s 500 x14 x15 x16 x17 x18 x19 x20 x21 x22 x23 x24 x25 x26 𝜶 𝒑 -0.0001 0.0020 0.0032* 0.0012 0.0010 -0.0000 0.0015 0.0008 0.0034 0.0013 0.0010 0.0014 0.0036* 𝜷 𝒑 ∗ 𝒓𝒎 0.9260*** 1.0084*** 1.0460*** 0.9762*** 0.8734*** 0.9901*** 1.0260*** 0.9527*** 0.9567*** 0.9565*** 0.9877*** 0.8552*** 0.8895*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 0.3840 -0.0804 0.4341 -0.0622 -0.6298* -0.2703 -0.8943** 0.5889 -0.3531 0.3844 0.3101 0.1527 -0.3686 𝜷 𝑺𝑴𝑩 -0.0371 0.1525*** 0.1246** 0.0872*** 0.0446 1.0168*** 0.5512*** 0.4563*** 0.8428*** 0.4065*** 0.4998*** 0.9132*** 0.3470*** 𝜷 𝑺𝑴𝑩𝟐 0.8889 0.1923 0.9805 -2.3540*** -0.0398 0.7309 -2.5226 -1.5504 -1.4545 -0.0745 -1.3886 1.9880 -4.0749** 𝜷 𝑯𝑴𝑳 0.0001 -0.3465*** -0.2609*** 0.1959*** 0.0773** 0.2825*** 0.1227** -0.1802*** -0.4774*** -0.1948*** -0.3068*** 0.3039*** -0.2078* 𝜷 𝑯𝑴𝑳𝟐 1.0001 -0.8814 -2.2776** 0.2735 -0.1539 0.1650 1.1621 -1.1825 -1.2956 -0.1121 -1.5983 -5.0345** 2.1665 𝜷 𝑹𝑴𝑾 0.0410 -0.1839** -0.2854*** -0.0966** 0.0919** 0.1404*** 0.0170 -0.0482 -0.2950*** -0.1099 -0.2161** 0.3190 0.2364* 𝜷 𝑹𝑴𝑾𝟐 -1.8945 -5.2273 -10.2594*** 0.3956 0.9817 -2.5885 -5.1079** -5.6789** -9.2478** 0.3401 -10.7406*** -6.6597 -7.5534** 𝜷 𝑪𝑴𝑨 0.0885 -0.1987** -0.3551*** -0.0190 0.1606*** -0.0980 -0.1133 -0.2976*** -0.2444* -0.1754* -0.2503*** 0.0154 0.2893 𝜷 𝑪𝑴𝑨𝟐 -4.0043 1.5153 1.0369 0.7959 3.1673 3.6944 -0.7514 0.7255 7.9073 -0.1746 7.8624* 12.3859** 3.5548 𝑹 𝟐 adj. 0.9446 0.9255 0.9181 0.9686 0.9515 0.9702 0.9385 0.9338 0.8972 0.9575 0.9280 0.9185 0.8726 This table presents regression estimates for the US conventional funds, obtained by from the Treynor and Mazuy (1966) extended to a conditional multifactor setting regressions with S&P500 as benchmark, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability ( 𝜶 𝒑), systematic risk ( 𝜷 𝒑), factor loadings associated to size (SMB), book-to-market (HML), profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination ( 𝑅  adj) . rm2, SMB2, HML2, RMW2 and CMA2 refers to squared risk factors. The predetermined information variables are the short-term rate (ST) and the dividend yield (DY). The time-varying alphas and betas associated with the risk factors are omitted. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 107 This table presents regression estimates for the US conventional funds, obtained by from the Treynor and Mazuy (1966) extended to a conditional multifactor setting regressions with KLD400 as benchmark, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability (𝜶𝒑), systematic risk (𝜷𝒑), factor loadings associated to size (SMB), book-to-market (HML), profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination (𝑅 adj) . rm2, SMB2, HML2, RMW2 and CMA2 refers to squared risk factors. The predetermined information variables are the short-term rate (ST) and the dividend yield (DY). The time-varying alphas and betas associated with the risk factors are omitted. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). Appendix 16 - Selectivity and timing abilities - Conditional Fama and French (2015) - MSCI KLD 400 - Conventional funds MSCI KLD 400 x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 𝜶 𝒑 -0.0024 0.0023 0.0029 -0.0019 0.0002 -0.0040* 0.0021 0.0013 0.0048** 0.0010 -0.0008 0.0007 -0.0002 𝜷 𝒑 ∗ 𝒓𝒎 0.9019*** 1.0804*** 0.6386*** 0.9872*** 0.9644*** 0.9601*** 0.8720*** 0.9466*** 1.0341*** 1.0577*** 0.7809*** 1.0058*** 0.9255*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 0.1267 -1.0213* -1.9421 -1.4472** 0.0270 1.7015* -1.0547** -0.6706* -0.5854 -1.8368** -0.2004 -0.4351* 0.2698 𝜷 𝑺𝑴𝑩 0.0798* 0.2100*** -0.0476 0.0590 0.0264 0.1213** 0.2812*** -0.0393 0.2421*** 0.5300*** 0.1412*** 0.2539*** 0.1513*** 𝜷 𝑺𝑴𝑩𝟐 -0.1727 -0.0369 -3.1169 2.3432 -3.1855* 1.3919 0.0836 0.2053 -0.5080 -2.8327* 0.3193 -0.1795 -0.2351 𝜷 𝑯𝑴𝑳 -0.0222 -0.3241*** -0.2857** 0.0678 0.2153* -0.1118 0.2427*** 0.2042*** -0.2424*** -0.2816*** -0.0259 0.1178*** -0.0083 𝜷 𝑯𝑴𝑳𝟐 -1.5999* -3.9219*** 1.0124 -1.2284 -0.4200 -1.3507 0.7049 -1.7846** -3.5956*** -2.4108* -2.1134** -1.4531*** 0.5241 𝜷 𝑹𝑴𝑾 -0.1616** -0.3797*** 0.3495** 0.1234 -0.1729 -0.1644 0.0469 0.0050 -0.3006*** 0.0268 -0.0453 0.0384 -0.1130*** 𝜷 𝑹𝑴𝑾𝟐 0.5236 -4.7737 -2.7938 3.1817 0.7834 -1.6225 2.1925 -0.3742 -6.4467* -8.1421 -2.0685 -1.2081 2.2925 𝜷 𝑪𝑴𝑨 -0.1739** -0.4746*** 0.3757 -0.1558 -0.1131 -0.6351*** 0.0789 0.0317 -0.5623*** -0.2910* 0.0577 -0.0828* -0.1903*** 𝜷 𝑪𝑴𝑨𝟐 4.6140 11.9577*** 9.7087 -0.8857 6.6137 2.7596 1.8586 4.8753 8.3876** 11.0751* 8.5083*** 5.9423*** -3.8959 𝑹 𝟐 adj. 0.9366 0.9149 0.8415 0.9672 0.9049 0.9332 0.9321 0.9543 0.9012 0.8606 0.9372 0.9823 0.9601 108 Appendix 16 - Selectivity and timing abilities - Conditional Fama and French (2015) - MSCI KLD 400 - Conventional funds - continued MSCI KLD 400 x14 x15 x16 x17 x18 x19 x20 x21 x22 x23 x24 x25 x26 𝜶 𝒑 0.0004 0.0024 0.0039** 0.0021* 0.0017 0.0004 0.0019 0.0013 0.0042* 0.0019 0.0012 0.0018 0.0032 𝜷 𝒑 ∗ 𝒓𝒎 0.9087*** 1.0030*** 1.0279*** 0.9518*** 0.8581*** 0.9835*** 1.0120*** 0.9409*** 0.9305*** 0.9554*** 0.9860*** 0.8448*** 0.8838*** 𝜷 𝒑 ∗ 𝒓𝒎𝟐 0.0175 -0.5078 0.0875 -0.6464 -1.1258*** -0.8296* -1.1923** 0.1882 -0.7131 -0.0994 -0.1119 -0.6070 -0.7228 𝜷 𝑺𝑴𝑩 -0.0630 0.1221*** 0.0963* 0.0671** 0.0242 0.9882*** 0.5282*** 0.4296*** 0.8233*** 0.3612*** 0.4538*** 0.8903*** 0.3132*** 𝜷 𝑺𝑴𝑩𝟐 0.9381 0.2519 0.8341 -2.2908*** 0.0093 0.9094 -2.5170 -1.5562 -1.5723 0.2077 -1.1774 1.6479 -4.2347** 𝜷 𝑯𝑴𝑳 0.0559 -0.2871*** -0.1988*** 0.2489*** 0.1247*** 0.3371*** 0.1823*** -0.1255** -0.4215*** -0.1862*** -0.2970*** 0.3090*** -0.2052* 𝜷 𝑯𝑴𝑳𝟐 -0.1533 -1.9310* -3.3856*** -0.7716 -0.9997 -0.9302 0.0474 -2.2566** -2.2512 -1.4018 -2.7737** -5.0884** 2.1604 𝜷 𝑹𝑴𝑾 0.0644 -0.1481* -0.2545*** -0.0698 0.1218** 0.1748** 0.0528 -0.0190 -0.2673** -0.1354 -0.2358** 0.3257* 0.2448** 𝜷 𝑹𝑴𝑾𝟐 -1.9144 -5.2702 -10.3113*** 0.4227 0.9606 -2.4965 -5.2495** -5.6294** -9.4161** -0.0093 -10.6396*** -5.0796 -5.2930 𝜷 𝑪𝑴𝑨 0.0516 -0.2324** -0.3915*** -0.0396 0.1410** -0.1274* -0.1509* -0.3284*** -0.2718* -0.1395 -0.2226** 0.0085 0.2865* 𝜷 𝑪𝑴𝑨𝟐 0.0324 5.7390 4.9156 4.7012 6.5990** 8.1583* 3.5159 4.6336 11.2972* 4.2076 12.8470*** 15.0536** 6.0612 𝑹 𝟐 adj. 0.9272 0.9225 0.9082 0.9435 0.9335 0.9660 0.9277 0.9284 0.8830 0.9538 0.9227 0.9065 0.8792 This table presents regression estimates for the US conventional funds, obtained by from the Treynor and Mazuy (1966) extended to a conditional multifactor setting regressions with KLD400 as benchmark, from February 2004 - September 2019. It reports the alpha coefficient that represents stock-picking ability ( 𝜶 𝒑), systematic risk ( 𝜷 𝒑), factor loadings associated to size (SMB), book-to-market (HML), profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination ( 𝑅  adj) . rm2, SMB2, HML2, RMW2 and CMA2 refers to squared risk factors. The predetermined information variables are the short-term rate (ST) and the dividend yield (DY). The time-varying alphas and betas associated with the risk factors are omitted. Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 109 Appendix 17 - Performance estimates using the conditional the Carhart four-factor model with a dummy - Standard & Poor`s 500 - Green funds x1 x4 x5 x6 x7 x8 x9 x10 x11 x12 𝜶 𝒑 -0.0028*** -0.0003 -0.0005 0.0012 -0.0007* -0.0011** 0.0009 -0.0019* -0.0013 -0.0000 𝜶 𝑫 -0.0028 0.0100** -0.0038 -0.0021 0.0011 -0.0001 0.0040 -0.0081* -0.0119** 0.0104 𝜷 𝒑 1.0667*** 1.0914*** 1.0705*** 0.8517*** 0.9910*** 0.9909*** 0.8733*** 1.1818*** 1.2244*** 0.9223*** 𝜷 𝑫 -0.2443*** -0.1838** -0.2323** -0.0300 0.0134 -0.0763*** -0.0740 -0.3059*** -0.2501*** -0.3791 𝜷 𝑺𝑴𝑩 0.1642*** 0.1440** 0.2794*** 0.2391*** 0.1495*** 0.0266 -0.0019 0.3358*** 0.5527*** 0.7613*** 𝜷 𝑺𝑴𝑩 ∗ 𝑫 0.3811 0.2347 0.7730*** 0.2914 0.0899*** 0.1393** 0.4799*** 0.9303*** 0.7960** 0.0517 𝜷 𝑯𝑴𝑳 0.1226*** -0.0408 -0.0222 -0.1105** -0.0032 -0.0809*** -0.0363 0.1416** 0.1657*** 0.0452 𝜷 𝑯𝑴𝑳 ∗ 𝑫 0.0299 0.0188 -0.1733* -0.0139 -0.0257 0.1090*** 0.0223 -0.2566** -0.0368 0.2581 𝜷 𝑴𝑶𝑴 -0.2406*** -0.1434*** -0.0665 -0.0676* -0.0111 -0.0107 0.0274 -0.1827*** -0.1546*** 0.0295 𝜷 𝑴𝑶𝑴 ∗ 𝑫 0.1021** -0.0833 -0.1320** -0.0785 0.0125 -0.0144 0.0046 -0.0550 -0.1520** -0.0543 𝑹 𝟐 adj. 0.9396 0.9044 0.8791 0.9017 0.9886 0.9787 0.9256 0.9337 0.9300 0.9336 This table presents regression estimates for the US green funds, obtained by the regression of the four-factor model with a dummy for the S&P500 as benchmark, from February 2004 - September 2019. The dummy variable is added in order to distinguish recessio n from expansion periods. It reports for both periods, estimates of performance ( 𝛂 𝐩), the systematic risk ( 𝛃 𝐩), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination ( R  adj. ). Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 110 Appendix 17 - Performance estimates using the conditional the Carhart four-factor model with a dummy - MSCI KLD 400 – Green funds x1 x4 x5 x6 x7 x8 x9 x10 x11 x12 𝜶 𝒑 -0.0023** 0.0001 -0.0000 0.0016 -0.0002 -0.0008*** 0.0013** -0.0013 -0.0007 0.0002 𝜶 𝑫 -0.0040 0.0101** -0.0041 -0.0028 0.0004 -0.0001 0.0028 -0.0097* -0.0131** 0.0098 𝜷 𝒑 1.0602*** 1.0889*** 1.0656*** 0.8409*** 0.9849*** 1.0016*** 0.8694*** 1.1745*** 1.2098*** 0.9215*** 𝜷 𝑫 -0.2016* -0.0977 -0.1615 0.0294 0.0835** -0.0075** -0.0408 -0.2722** -0.1836* -0.3802 𝜷 𝑺𝑴𝑩 0.1242** 0.1020* 0.2382*** 0.2107*** 0.1122*** -0.0194*** -0.0356 0.2914*** 0.5102*** 0.7243*** 𝜷 𝑺𝑴𝑩 ∗ 𝑫 0.3072 0.1152 0.6742*** 0.1959 -0.0281 0.0263*** 0.4077*** 0.8635*** 0.6961** -0.0003 𝜷 𝑯𝑴𝑳 0.1664*** 0.0184 0.0460 -0.0649 0.0598*** -0.0164** 0.0194 0.2168*** 0.2430*** 0.0708 𝜷 𝑯𝑴𝑳 ∗ 𝑫 -0.0402 -0.0739 -0.2713** -0.0870 -0.1229*** 0.0112* -0.0583 -0.3583*** -0.1460 0.2468 𝜷 𝑴𝑶𝑴 -0.2346*** -0.1246*** -0.0408 -0.0548 0.0124 0.0162*** 0.0484** -0.1547*** -0.1271*** 0.0291 𝜷 𝑴𝑶𝑴 ∗ 𝑫 0.1066** -0.0766 -0.1377** -0.0769 0.0082 -0.0171*** -0.0081 -0.0757 -0.1642** -0.0573 𝑹 𝟐 adj. 0.9299 0.9080 0.8790 0.8905 0.9830 0.9985 0.9159 0.9266 0.9221 0.9288 This table presents regression estimates for the US green funds, obtained by the regression of the four-factor model with a dummy for the KLD400 as benchmark, from February 2004 - September 2019. The dummy variable is added in order to distinguish recessio n from expansion periods. It reports for both periods, estimates of performance ( 𝛂 𝐩 ), the systematic risk ( 𝛃 𝐩 ), factor loadings associated to size (SMB), book-to-market (HML) and momentum (MOM) factors and the adjusted coefficient of determination ( R  adj. ). Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 117 Appendix 20 - Performance estimates using the conditional the Fama and French (2015) five-factor model with a dummy - Standard & Poor`s 500 - Conventional funds x1 x2 x7 x8 x9 x10 x11 x12 x13 x14 𝜶 𝒑 -0.0021*** 0.0014 0.0012 -0.0002 0.0030*** -0.0012 -0.0006 -0.0001 -0.0003 0.0003 𝜶 𝑫 0.0006 -0.0207*** 0.0082** 0.0029 -0.0161*** -0.0281*** -0.0054*** -0.0000 0.0025 0.0016 𝜷 𝒑 0.9038*** 1.0860*** 0.8635*** 0.9584*** 1.0623*** 1.0580*** 0.7680*** 1.0257*** 0.9320*** 0.9075*** 𝜷 𝑫 0.1220** -0.0876 0.0432 0.0651** -0.1035 -0.1285 0.2587*** -0.0022 0.0056 0.2092** 𝜷 𝑺𝑴𝑩 0.0915** 0.2291*** 0.2769*** -0.0171 0.2365*** 0.5272*** 0.1278*** 0.2755*** 0.1545*** -0.0206 𝜷 𝑺𝑴𝑩 ∗ 𝑫 0.1318 0.3412** 0.1941 -0.1115 0.4254*** 0.2777 -0.0022 0.0424 0.1494 -0.1137 𝜷 𝑯𝑴𝑳 -0.0415 -0.3249*** 0.2107*** 0.1176*** -0.2910*** -0.2693*** -0.0566 0.0672*** -0.0755** -0.0108 𝜷 𝑯𝑴𝑳 ∗ 𝑫 -0.2031*** -0.0302 -0.2120** -0.0402 -0.1434* -0.5354*** -0.0626 -0.0753** -0.0012 0.1377 𝜷 𝑹𝑴𝑾 -0.1617*** -0.4209*** 0.0447 -0.0232 -0.4321*** -0.1404 -0.0873 0.0142 -0.1426*** 0.0081 𝜷 𝑹𝑴𝑾 ∗ 𝑫 0.3133 1.0231*** -0.4233** 0.2024* 0.8238*** 0.9740*** 0.2243 0.0174 -0.0545 0.1674 𝜷 𝑪𝑴𝑨 -0.1094* -0.4957*** 0.0619 0.0343 -0.4869*** -0.3601*** 0.1282 -0.0411* -0.1066** 0.0920 𝜷 𝑪𝑴𝑨 ∗ 𝑫 -0.4104** -0.9438*** 0.8254*** -0.0924 -0.8403*** -0.7776** -0.7837*** -0.0359 0.1412 -0.2120 𝑹 𝟐 adj. 0.9579 0.9302 0.9376 0.9681 0.9105 0.8739 0.9492 0.9948 0.9583 0.9453 This table presents regression estimates for the US conventional funds, obtained by the regression of the five-factor model with a dummy for the S&P500 as benchmark, from February 2004 - September 2019.The dummy variable is added in order to distinguish recessions from expansions periods. It reports for both periods, estimates of performance ( 𝛂 𝐩), the systematic risk ( 𝛃 𝐩), factor loadings associated to size (SMB), book-to-market (HML) and profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination ( R  adj.). Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 118 Appendix 20 - Performance estimates using the conditional the Fama and French (2015) five-factor model with a dummy - Standard & Poor`s 500 - Conventional funds - continued x15 x16 x17 x18 x19 x20 x21 x22 x23 x24 𝜶 𝒑 0.0009 0.0017 -0.0006 0.0011** 0.0005 -0.0003 -0.0006 0.0019 0.0020* 0.0003 𝜶 𝑫 -0.0049 -0.0066** 0.0084*** -0.0012 -0.0038 -0.0050 -0.0070* -0.0193*** -0.0158*** -0.0157*** 𝜷 𝒑 1.0304*** 1.0493*** 0.9963*** 0.8654*** 0.9801*** 1.0056*** 0.9923*** 0.9603*** 0.9567*** 1.0016*** 𝜷 𝑫 -0.0940 -0.0584 -0.1287*** -0.0274 0.2381*** 0.1914** -0.0941 -0.2196** 0.1145** -0.1460** 𝜷 𝑺𝑴𝑩 0.1694*** 0.1114* 0.0363 0.0405 1.0019*** 0.4930*** 0.4581*** 0.8228*** 0.3950*** 0.5494*** 𝜷 𝑺𝑴𝑩 ∗ 𝑫 0.1119 0.3116** -0.0624 -0.0703 0.1233 0.4058* -0.0047 0.1535 0.3519*** -0.0960 𝜷 𝑯𝑴𝑳 -0.3574*** -0.2661*** 0.1912*** 0.1074*** 0.2779*** 0.2011*** -0.2016*** -0.4836*** -0.1784*** -0.3009*** 𝜷 𝑯𝑴𝑳 ∗ 𝑫 -0.0640 -0.2542** -0.1616*** -0.0689 -0.1868** -0.7963*** 0.0082 0.1320 -0.1477 -0.1049 𝜷 𝑹𝑴𝑾 -0.2019** -0.3636*** -0.0787* 0.1310*** 0.1137** -0.0516 -0.1589*** -0.3835*** -0.1553* -0.2007** 𝜷 𝑹𝑴𝑾 ∗ 𝑫 -0.0811 0.3438** -0.2789** -0.2251 0.5754** -0.0819 0.3380* 0.6456** 1.4601*** 0.4903* 𝜷 𝑪𝑴𝑨 -0.1668* -0.2991** -0.0191 0.1001** -0.0870 -0.1662** -0.2716*** -0.2531* -0.0795 -0.2822*** 𝜷 𝑪𝑴𝑨 ∗ 𝑫 -0.5131 -0.7711*** 0.0559 0.3663*** -0.2676 -0.3850 -0.7189*** -1.2395*** -2.2057*** -0.7040*** 𝑹 𝟐 adj. 0.9269 0.9124 0.9668 0.9499 0.9699 0.9382 0.9292 0.8915 0.9531 0.9276 This table presents regression estimates for the US conventional funds, obtained by the regression of the five-factor model with a dummy for the S&P500 as benchmark, from February 2004 - September 2019. The dummy variable is added in order to distinguish recessions from expansions periods. It reports for both periods, estimates of performance ( 𝛂 𝐩), the systematic risk ( 𝛃 𝐩), factor loadings associated to size (SMB), book-to-market (HML) and profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination ( R  adj.). Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 119 Appendix 20 - Performance estimates using the conditional the Fama and French (2015) five-factor model with a dummy - MSCI KLD 400 - Conventional funds x1 x2 x7 x8 x9 x10 x11 x12 x13 x14 𝜶 𝒑 -0.0016* 0.0019 0.0017* 0.0003 0.0036*** -0.0005 -0.0002 0.0005 0.0001 0.0009 𝜶 𝑫 -0.0037 -0.0252*** 0.0044 -0.0015 -0.0205*** -0.0325*** -0.0098*** -0.0045* -0.0012 -0.0030 𝜷 𝒑 0.8940*** 1.0786*** 0.8553*** 0.9363*** 1.0440*** 1.0298*** 0.7581*** 1.0102*** 0.9256*** 0.8862*** 𝜷 𝑫 0.1480 -0.1010 0.0734 0.1109* -0.0815 -0.1049 0.2828*** 0.0412 0.0554 0.2742*** 𝜷 𝑺𝑴𝑩 0.0565 0.1838*** 0.2493*** -0.0437 0.2071*** 0.5018*** 0.1033** 0.2431*** 0.1250*** -0.0435 𝜷 𝑺𝑴𝑩 ∗ 𝑫 0.0480 0.2975 0.1111 -0.2090 0.3521** 0.2095 -0.0951 -0.0523 0.0495 -0.2392** 𝜷 𝑯𝑴𝑳 -0.0063 -0.2853*** 0.2724*** 0.1823*** -0.2128*** -0.1909** -0.0100 0.1305*** -0.0075 0.0563 𝜷 𝑯𝑴𝑳 ∗ 𝑫 -0.2311** -0.0451 -0.2709*** -0.1013 -0.2093* -0.5980*** -0.1009 -0.1370** -0.0763 0.0659 𝜷 𝑹𝑴𝑾 -0.1776*** -0.4402*** 0.0628 -0.0138 -0.4154*** -0.1282 -0.0888 0.0117 -0.1238*** 0.0199 𝜷 𝑹𝑴𝑾 ∗ 𝑫 0.5122* 1.1759*** -0.2703 0.3846*** 0.9643*** 1.1042*** 0.4064* 0.2167** 0.1288 0.3867** 𝜷 𝑪𝑴𝑨 -0.1301 -0.5136*** 0.0119 -0.0194 -0.5535*** -0.4274*** 0.0848 -0.0990** -0.1641*** 0.0346 𝜷 𝑪𝑴𝑨 ∗ 𝑫 -0.7408*** -1.3178*** 0.5756*** -0.3786*** -1.1177*** -1.0549*** -1.0945*** -0.3122*** -0.0828 -0.5009*** 𝑹 𝟐 adj. 0.9399 0.9141 0.9314 0.9453 0.8970 0.8528 0.9344 0.9812 0.9587 0.9282 This table presents regression estimates for the US conventional funds, obtained by the regression of the five-factor model with a dummy for the KLD400 as benchmark, from February 2004 - September 2019.The dummy variable is added in order to distinguish recessions from expansions periods. It reports for both periods, estimates of performance ( 𝛂 𝐩), the systematic risk ( 𝛃 𝐩), factor loadings associated to size (SMB), book-to-market (HML) and profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination ( R  adj.). Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*). 120 Appendix 20 - Performance estimates using the conditional the Fama and French (2015) five-factor model with a dummy - MSCI KLD 400 - Conventional funds - continued x15 x16 x17 x18 x19 x20 x21 x22 x23 x24 𝜶 𝒑 0.0014 0.0023* 0.0001 0.0016** 0.0011 0.0003 -0.0001 0.0026 0.0022* 0.0005 𝜶 𝑫 -0.0088** -0.0107*** 0.0043 -0.0049* -0.0088* -0.0102 -0.0106** -0.0229*** -0.0216*** -0.0202*** 𝜷 𝒑 1.0218*** 1.0318*** 0.9675*** 0.8436*** 0.9673*** 0.9825*** 0.9775*** 0.9354*** 0.9588*** 1.0025*** 𝜷 𝑫 -0.0570 0.0039 -0.0951 0.0132 0.2852*** 0.2351** -0.0327 -0.2025* 0.2194*** -0.0918 𝜷 𝑺𝑴𝑩 0.1373*** 0.0822 0.0133 0.0193 0.9732*** 0.4675*** 0.4297*** 0.7995*** 0.3511*** 0.5040*** 𝜷 𝑺𝑴𝑩 ∗ 𝑫 0.0255 0.2048 -0.1335 -0.1502 -0.0001 0.2914 -0.1041 0.1052 0.2711** -0.1408 𝜷 𝑯𝑴𝑳 -0.2822*** -0.1889** 0.2652*** 0.1715*** 0.3497*** 0.2755*** -0.1288** -0.4125*** -0.1583*** -0.2795*** 𝜷 𝑯𝑴𝑳 ∗ 𝑫 -0.1392 -0.3383*** -0.2252*** -0.1297** -0.2573** -0.8630*** -0.0738 0.0754 -0.2417** -0.1621* 𝜷 𝑹𝑴𝑾 -0.1818** -0.3470*** -0.0682 0.1415*** 0.1308** -0.0383 -0.1423** -0.3721*** -0.1733** -0.2200** 𝜷 𝑹𝑴𝑾 ∗ 𝑫 0.0824 0.5393*** -0.1454 -0.0794 0.7939*** 0.1204 0.5212*** 0.7426** 1.7805*** 0.7152** 𝜷 𝑪𝑴𝑨 -0.2305** -0.3648*** -0.0826 0.0452 -0.1480** -0.2298*** -0.3335*** -0.3141** -0.0670 -0.2697*** 𝜷 𝑪𝑴𝑨 ∗ 𝑫 -0.7512* -1.0047*** -0.1896 0.1419 -0.6029*** -0.7288*** -0.9198*** -1.4589*** -2.4315*** -0.9434*** 𝑹 𝟐 adj. 0.9237 0.9035 0.9368 0.9257 0.9641 0.9222 0.9232 0.8765 0.9500 0.9243 This table presents regression estimates for the US conventional funds, obtained by the regression of the five-factor model with a dummy for the KLD400 as benchmark, from February 2004 - September 2019.The dummy variable is added in order to distinguish recessions from expansions periods. It reports for both periods, estimates of performance ( 𝛂 𝐩), the systematic risk ( 𝛃 𝐩), factor loadings associated to size (SMB), book-to-market (HML) and profitability (RMW) and investment (CMA) factors and the adjusted coefficient of determination ( R  adj.). Standard errors are corrected for autocorrelation and heteroscedasticity following Newey and West (1987). The asterisks are used to identify statistical significance of the coefficients to a level of significance of 1% (***), 5% (**) and 10% (*).